Cics Db2 Guide
Cics Db2 Guide
SC34-6252-00
CICS Transaction Server for z/OS
SC34-6252-00
Note!
Before using this information and the product it supports, be sure to read the general information under “Notices” on page
213.
Summary of changes . . . . . . . . . . . . . . . . . . . . . . xi
Changes for CICS Transaction Server for z/OS, Version 2 Release 3 . . . . . xi
Changes for CICS Transaction Server for z/OS, Version 2 Release 2 . . . . . xi
Changes for CICS Transaction Server for z/OS, Version 2 Release 1 . . . . . xiii
Earlier releases . . . . . . . . . . . . . . . . . . . . . . . . xiv
Contents v
SQL, threadsafe and other programming considerations for CICS DB2
applications . . . . . . . . . . . . . . . . . . . . . . . . . 105
Enabling CICS DB2 applications to exploit the open transaction environment
(OTE) through threadsafe programming . . . . . . . . . . . . . 106
SQL language . . . . . . . . . . . . . . . . . . . . . . . . 108
Using qualified and unqualified SQL . . . . . . . . . . . . . . . 109
Views . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
Updating index columns . . . . . . . . . . . . . . . . . . . . 110
Dependency of unique indexes . . . . . . . . . . . . . . . . . 110
Commit processing . . . . . . . . . . . . . . . . . . . . . . 110
Serializing transactions . . . . . . . . . . . . . . . . . . . . 111
Page contention . . . . . . . . . . . . . . . . . . . . . . . 112
CICS and CURSOR WITH HOLD option . . . . . . . . . . . . . . 113
EXEC CICS RETURN IMMEDIATE command . . . . . . . . . . . . 114
Avoiding AEY9 abends . . . . . . . . . . . . . . . . . . . . 114
Chapter 8. Using JDBC and SQLJ to access DB2 data from Java
programs and enterprise beans written for CICS . . . . . . . . . . 117
Making JDBC and SQLJ work in the CICS DB2 environment . . . . . . . 117
Requirements to support Java programs in the CICS DB2 environment . . . . 119
Programming with JDBC and SQLJ in the CICS DB2 environment . . . . . 122
Acquiring a connection to a database . . . . . . . . . . . . . . . . 122
How many connections can you have? . . . . . . . . . . . . . . 123
Acquiring a connection using the JDBC DriverManager interface . . . . . . 123
Acquiring a connection using the DataSource interface . . . . . . . . . . 124
Setting up the sample applications to publish, look up and retract a
DataSource . . . . . . . . . . . . . . . . . . . . . . . . 125
Publishing a DataSource using CICSDataSourcePublish.java . . . . . . 126
Looking up a DataSource using CICSjdbcDataSource.java . . . . . . . 127
Retracting a DataSource using CICSDataSourceRetract.java . . . . . . 128
Committing a unit of work . . . . . . . . . . . . . . . . . . . . 129
Autocommit. . . . . . . . . . . . . . . . . . . . . . . . . 129
Syncpoint issues for explicit and default URLs . . . . . . . . . . . . 130
CICS abends during JDBC or SQLJ requests . . . . . . . . . . . . . 130
Using JDBC and SQLJ in enterprise beans: special considerations . . . . . 130
Chapter 9. Preparing CICS DB2 programs for execution and production 133
The CICS DB2 test environment . . . . . . . . . . . . . . . . . . 133
CICS DB2 program preparation steps . . . . . . . . . . . . . . . . 134
CICS SQLCA formatting routine . . . . . . . . . . . . . . . . . 136
What to bind after a program change . . . . . . . . . . . . . . . . 137
Bind options and considerations for programs . . . . . . . . . . . . . 138
RETAIN . . . . . . . . . . . . . . . . . . . . . . . . . . 138
Isolation level . . . . . . . . . . . . . . . . . . . . . . . . 139
Plan validation time . . . . . . . . . . . . . . . . . . . . . . 139
ACQUIRE and RELEASE . . . . . . . . . . . . . . . . . . . 139
CICS DB2 program testing and debugging . . . . . . . . . . . . . . 139
Going into production: checklist for CICS DB2 applications . . . . . . . . 139
Tuning a CICS application that accesses DB2 . . . . . . . . . . . . . 142
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . 201
The CICS Transaction Server for z/OS library . . . . . . . . . . . . . 201
The entitlement set . . . . . . . . . . . . . . . . . . . . . . 201
PDF-only books . . . . . . . . . . . . . . . . . . . . . . . 201
Other CICS books . . . . . . . . . . . . . . . . . . . . . . . 203
Books from related libraries . . . . . . . . . . . . . . . . . . . . 203
DB2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
DB2 Performance Monitor (DB2 PM) . . . . . . . . . . . . . . . 204
Contents vii
Resource Management Facility (RMF) . . . . . . . . . . . . . . . 204
Determining if a publication is current . . . . . . . . . . . . . . . . 204
Accessibility . . . . . . . . . . . . . . . . . . . . . . . . . 205
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
Notices . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
Programming interface information . . . . . . . . . . . . . . . . . 214
Trademarks. . . . . . . . . . . . . . . . . . . . . . . . . . 214
The aim of this book is to give introductory and guidance information on evaluating,
installing, and using the CICS DB2 attachment facility, and on defining and
maintaining your CICS DB2 environment.
Notes on terminology
When the term “CICS” is used without any qualification, it refers to the CICS
element of IBM® CICS Transaction Server for z/OS.
“DB2” without any qualification refers to DB2 Universal Database Server for OS/390
and z/OS.
“MVS™” is used for the operating system, which can be an element of z/OS,
OS/390, or MVS/Enterprise System Architecture System Product (MVS/ESA™ SP).
This part lists briefly the changes that have been made for the following recent
releases:
Technical changes
v There are changes to the samples that are provided to set up a
CICS-compatible DataSource. Chapter 8, “Using JDBC and SQLJ to access DB2
data from Java programs and enterprise beans written for CICS,” on page 117
has information on using the DataSource interface to acquire connections.
Structural changes
v Chapter 1, “Overview of the CICS DB2 interface,” on page 1, has been rewritten,
and now includes information on
– how threads and thread TCBs work
– how the DB2CONN, DB2ENTRY and DB2TRAN definitions work together
– how plans, packages and dynamic plan exits work
v Chapter 7, “Application design and development considerations for CICS DB2,”
on page 87 and Chapter 9, “Preparing CICS DB2 programs for execution and
production,” on page 133 have been reorganised. In particular, Chapter 8, “Using
JDBC and SQLJ to access DB2 data from Java programs and enterprise beans
written for CICS,” on page 117 is now a separate chapter.
v Chapter 2, “Installation and migration notes for CICS DB2,” on page 15 and
Chapter 5, “Defining the CICS DB2 connection,” on page 51 have been
reorganised.
v The chapter ’Customization: dynamic plan exits’ has been removed, and the
information it contained is now in “Dynamic plan exits” on page 96 and “Dynamic
plan switching” on page 99.
v The chapter ’Monitoring, tuning and handling deadlocks’ has been removed. The
information it contained is now in Chapter 10, “Accounting and monitoring in a
CICS DB2 environment,” on page 145, Chapter 9, “Preparing CICS DB2
programs for execution and production,” on page 133, Chapter 11, “Problem
determination for CICS DB2,” on page 177, and the CICS Performance Guide.
For most items in each of the lists above, there is a reference to the part of the
book where there is more detail.
Earlier releases
Changes for CICS Transaction Server for OS/390 Version 1 Release 3
Major changes for this edition provide information on the changes to the INITPARM
system initialization parameter now that CICS no longer supports running the CICS
DB2 attachment facility with a macro RCT. See “Effect of migration to RDO on the
INITPARM system initialization parameter” on page 21 and “Migrating to RDO for
DB2 resource definition” on page 21.
The CICS DB2 attachment facility creates an overall connection between CICS and
DB2. CICS applications use this connection to issue commands and requests to
DB2. The connection between CICS and DB2 can be created or terminated at any
time, and CICS and DB2 can be started and stopped independently. You can name
an individual DB2 subsystem to which CICS connects, or (if you have DB2 Version
7 or later) you can use the group attach facility to let DB2 choose any active
member of a data-sharing group of DB2 subsystems for the connection. You also
have the option of CICS automatically connecting and reconnecting to DB2. A DB2
system can be shared by several CICS systems, but each CICS system can be
connected to only one DB2 subsystem at a time.
Attachment commands display and control the status of the CICS DB2 attachment
facility, and are issued using the CICS supplied transaction DSNC. The attachment
commands are:
v STRT - start the connection to DB2
v STOP - stop the connection to DB2
v DISP - display the status of threads, and display statistics
v MODI - modify characteristics of the connection to DB2
v DISC - disconnect threads
The connection between CICS and DB2 is a multithread connection. Within the
overall connection between CICS and DB2, there is a thread—an individual
connection into DB2—for each active CICS transaction accessing DB2. Threads
allow each CICS transaction to access DB2 resources, such as a command
processor or an application plan (the information that tells DB2 what the application
program’s SQL requests are, and the most efficient way to service them). See
“Overview: How threads work” on page 2 below for a full explanation of how
threads work.
When an application program operating in the CICS environment issues its first
SQL request, CICS and DB2 process the request as follows:
v A language interface, or stub, DSNCLI, that is link-edited with the application
program calls the CICS resource manager interface (RMI).
v The RMI processes the request, and passes control to the CICS DB2 attachment
facility’s task-related user exit (TRUE), the module that invokes DB2 for each
task.
CICS DSN1MSTR
address space address space
DSN1DBM1
address space
DSN1DIST
address space
IRLMPROC
address space
DSN1SPAS
address space
There are different types of thread, and you can set a limit on the number of each
type of thread that can be active at any one time. This prevents the overall CICS
DB2 connection from becoming overloaded with work. A special type of thread is
used for DB2 commands issued using the DSNC transaction, and you can also
define special threads for CICS transactions with particular requirements, such as
transactions that require a fast response time. You can define what a transaction
must do if no more threads of the type it needs are available — it can wait until a
thread of the right type is available; it can use a general-purpose thread, called a
pool thread; or it can abend.
The types of thread provided by the CICS DB2 attachment facility are:
Command threads
Command threads are reserved by the CICS DB2 attachment facility for
issuing commands to DB2 using the DSNC transaction. They are not used
for commands acting on the CICS DB2 attachment facility itself, because
these commands are not passed to DB2. When a command thread is not
available, commands automatically overflow to the pool, and use a pool
thread. Command threads are defined in the command threads section of
the DB2CONN definition.
Entry threads
Entry threads are specially defined threads intended for transactions with
special requirements, such as transactions that require a fast response
time, or transactions with special accounting needs. You can instruct the
CICS DB2 attachment facility to give entry threads to particular CICS
transactions. You define the different types of entry threads that are needed
for different transactions, and you can set a limit on the number of each of
these types of entry thread. If a transaction is permitted to use an entry
thread, but no suitable entry thread is available, the transaction can
overflow to the pool and use a pool thread, or wait for a suitable entry
thread, or abend, as you have chosen in the definition for the entry thread.
A certain number of each type of entry thread can be protected. When an
entry thread is released, if it is protected it is not terminated immediately. It
is kept for a period of time, and if another CICS transaction needs the same
type of entry thread during that period, it is reused. This avoids the
overhead involved in creating and terminating the thread for each
transaction. An entry thread that is unprotected is terminated immediately,
unless a CICS transaction is waiting to use it the moment it is released.
Entry threads are defined using a DB2ENTRY definition.
Pool threads
Pool threads are used for all transactions and commands that are not using
an entry thread or a DB2 command thread. Pool threads are intended for
low volume transactions, and for overflow transactions that could not obtain
For more detailed information on how the different types of thread are created, used
and terminated, see “How threads are created, used, and terminated” on page 55.
Each thread runs under a thread task control block (thread TCB) that belongs to
CICS. CICS and DB2 both have connection control blocks linked to the thread TCB.
They use these connection control blocks to manage the thread into DB2, and to
communicate information to each other about the thread. The DB2 connection
control block controls the thread within DB2. The CICS connection control block,
called the CSUB, acts as a pointer to the DB2 connection control block, and
contains the information CICS requires to call the DB2 connection control block
when the thread is needed. DB2 calls these connection control blocks “agent
structures”.
The nature of the thread TCBs, and the way in which they are linked to the DB2
connection control block (and therefore to the thread), differs depending on the
version of DB2 to which CICS is connected.
While CICS is connecting to a DB2 subsystem, it checks the DB2 release level of
the subsystem. If CICS is connecting to DB2 Version 6 or later, the CICS DB2
task-related user exit (the module that invokes DB2 for each task) is automatically
enabled as open API, so it can use the open transaction environment (OTE). If
CICS is connecting to DB2 Version 5 or earlier, the task-related user exit is not
enabled as open API, and does not use the open transaction environment.
Once created, the subtask thread TCBs are permanently associated with a
particular CSUB and DB2 connection control block. When the process for which a
subtask thread TCB was created is complete, the subtask thread TCB, CSUB, DB2
connection control block and thread are released, and the whole assembly can be
reused by another CICS transaction to access DB2 resources. So to reuse an
existing thread, the CICS DB2 attachment facility must also reuse the subtask
thread TCB associated with it.
If the thread is terminated before it is reused, the subtask thread TCB and its
associated CSUB and DB2 connection control block remain available in the system.
Like the threads themselves, it takes processor resource to create these, so the
CICS DB2 attachment facility reuses them. If a thread is requested and no existing
threads are available, the CICS DB2 attachment facility looks for an unused subtask
thread TCB, CSUB and DB2 connection control block, and reuses them to run a
new thread into DB2.
Resource Manager
Interface (RMI)
1
CICS DB2
TRUE
Subtask
TCB DB2
DB2 connection thread
CSUB Plan
control block
Subtask
TCB
2
DB2 connection thread
CSUB Plan
control block
Subtask
TCB
3
DB2 connection
CSUB
control block
In Figure 2, situation 1 shows CICS using a thread to access DB2. On the CICS
main TCB, the language interface that is link-edited with the application program
calls the Resource Manager Interface (RMI), which invokes the CICS DB2
attachment facility’s task-related user exit. The CICS DB2 task-related user exit,
operating on the CICS main TCB, uses an assembly consisting of a subtask TCB, a
CSUB, and a DB2 connection control block to run a thread into DB2. The plan
associated with the thread is held in DB2.
Situation 2 shows a thread that is not currently in use, but is protected. The
subtask TCB, CSUB and DB2 connection control block are still assembled together,
and the thread runs into DB2. The thread is available for reuse.
Open TCBs are not permanently associated with a CSUB and DB2 connection
control block. The CICS DB2 attachment facility can associate them with any CSUB
and DB2 connection control block that are available, and therefore with any thread
that is available. When the open TCB no longer needs the connection to DB2, it
dissociates from, or releases, the thread, CSUB and DB2 connection control block.
The thread, CSUB and DB2 connection control block can then be used by a
different open TCB for the tasks it wishes to perform in DB2.
If the thread is terminated before it is reused, the CSUB and DB2 connection
control block remain available in the system. If no existing threads are available,
and an open TCB needs a connection to DB2, the CICS DB2 attachment facility
can associate the unused CSUB and DB2 connection control block with the open
TCB, and reuse them to run a new thread into DB2.
Figure 3 on page 7 summarizes how thread TCBs operate in the open transaction
environment.
Open TCB
CICS DB2
TRUE
1 DB2
DB2 connection thread
CSUB Plan
control block
DB2 connection
3 CSUB
control block
Open
TCB
4
Open
TCB
In Figure 3, situation 1 shows CICS using a thread to access DB2 in the open
transaction environment. The CICS DB2 task-related user exit has been invoked by
the Resource Manager Interface (RMI), and it is operating on an open TCB. The
CICS DB2 attachment facility has associated a CSUB and a DB2 connection control
block with the open TCB. The DB2 connection control block has a thread into DB2.
The plan associated with the thread is held in DB2.
Situation 2 shows a thread that is not currently in use, but is protected. The
CSUB and DB2 connection control block are still linked to each other and have a
thread, but no open TCB is attached to them. The thread is available for reuse.
Situation 3 shows an assembly that is left after a thread was terminated. The
CSUB and DB2 connection control block are available for reuse. They need a new
thread.
Situation 4 shows open TCBs that are available for reuse. The CICS DB2
attachment facility can use these open TCBs and associate CSUB and DB2
connection control block assemblies with them to run threads into DB2.
Both types of TCB that the CICS DB2 attachment facility uses to run the threads,
the open TCBs and the subtask TCBs, are referred to in this documentation as
“thread TCBs”. In many situations, the different nature of the two types of thread
TCB does not lead to any differences in the operation of the CICS DB2 connection.
As we have read in “Overview: How CICS connects to DB2” on page 1, the CICS
DB2 connection consists of an overall connection between CICS and DB2, and
individual connections known as threads. You can define the attributes of the overall
connection, and the attributes of the different types of thread. If you have specially
defined entry threads for key transactions, you can tell the CICS DB2 attachment
facility which CICS transactions can use those threads. Defining the CICS DB2
connection involves three different objects: DB2CONN (the DB2 connection
definition), DB2ENTRY (the DB2 entry definition), and DB2TRAN (the DB2
transaction definition). The scope of each object is as follows:
DB2CONN
DB2CONN is the main definition for the CICS DB2 connection. You need to
install a DB2CONN before you can start the CICS DB2 connection. In the
DB2CONN definition, you:
v Define the attributes of the overall CICS DB2 connection, which are:
– the DB2 subsystem that CICS connects to, or the data-sharing group of
DB2 subsystems from which DB2 picks an active member for CICS to
connect to
– if the connection to DB2 fails, whether CICS reconnects automatically or
not, and, if you are using the group attach facility, whether CICS
reconnects to the same DB2 subsystem or not
– a general-purpose authorization ID for the CICS DB2 attachment facility to
sign on to a thread, if no other authorization is needed for the thread
– the limit on the total number of TCBs that CICS can use to run threads
into DB2 at any one time
– how long protected threads are kept before termination
– how error messages are communicated to CICS, and what transactions
should do if their thread fails
– where messages and statistics are sent
v Define the attributes of command threads (the special threads used by the
DSNC transaction for issuing DB2 commands), which are:
– the limit on the number of command threads that CICS can use at any
one time
– what type of authorization ID DB2 checks when a command thread is
requested (for example, the ID of the user, the ID of the transaction, the
general-purpose CICS DB2 attachment facility ID)
v Define the attributes of pool threads (the general-purpose threads used when
transactions do not need a special command or entry thread, or when there
are no special threads left for a transaction to use), which are:
You can start the CICS DB2 connection with only a DB2CONN installed—you
do not need any DB2ENTRY and DB2TRAN definitions to make the connection.
If you do this, there are no special threads for key transactions (entry threads).
All transactions use general-purpose threads from the pool, and the most
important transactions have to wait just as long as the least important
transactions to get their individual connection into DB2. To ensure that your
important transactions are prioritized, create DB2ENTRY and, if necessary,
DB2TRAN definitions for them.
DB2ENTRY
You can set up many DB2ENTRY definitions to define different types of entry
threads. The entry threads can be used by the transactions that you specify, to
gain priority access (or specialized access) to DB2 resources. In effect, you are
reserving a certain number of threads that can only be used by those
transactions. You can also protect a number of each type of entry thread, which
improves performance for heavily-used transactions. In a DB2ENTRY definition,
you can specify a particular transaction, or (by using a wildcard) a group of
transactions, that are associated with the DB2ENTRY and can use the type of
entry thread that it defines. When those transactions request a thread, the CICS
DB2 attachment facility gives them that type of entry thread, if one is available.
If you want other transactions to use the same type of entry thread, you can
create a DB2TRAN definition for those transactions, which simply tells CICS
that the transactions are associated with a particular DB2ENTRY.
In each DB2ENTRY definition, you:
v Specify a transaction, or (by using a wildcard in the name) a group of
transactions, that can use this type of entry thread
v Define the attributes of this type of entry thread, which are:
– what type of authorization ID DB2 checks when this type of entry thread is
requested (for example, the ID of the user, the ID of the transaction, the
general-purpose CICS DB2 attachment facility ID)
– what the priority of the thread TCBs is relative to the CICS main TCB
– the limit on the number of this type of entry thread that CICS can use at
any one time
You can define and install DB2CONN, DB2ENTRY and DB2TRAN objects using
RDO. The objects can also be defined in batch using DFHCSDUP. CICSplex SM
uses EXEC CICS CREATE to install these objects. For detailed information about
how to define each of these objects, see the CICS Resource Definition Guide.
The plans are held within DB2, and each thread into DB2 relates to a plan. The
plan that each type of thread uses is named on the DB2CONN definition (for pool
threads) or the thread’s DB2ENTRY definition (for entry threads). The plan for a
particular type of thread must contain the bound form of the SQL statements from
all the application programs that use that type of thread to access DB2. You can
either name the plan explicitly, or name a dynamic plan exit, a routine that
determines which plan to use for the transaction that has requested a thread of that
type.
The first step is to put the program through the DB2 precompiler. The DB2
precompiler builds a database request model (DBRM) that contains information
about each of the program’s SQL statements.
The second, third and fourth steps are the normal process for preparing any CICS
application program, whether or not it accesses DB2. The second step is to put the
program through the CICS command language translator. The third step is to
compile or assemble the program. The fourth step is to link-edit the program with
the necessary interfaces (including the CICS DB2 language interface module
DSNCLI). The end product of Steps 2, 3 and 4 is an application load module that
enables the program to run. For more information on these steps, see “CICS DB2
program preparation steps” on page 134.
An extra step is required to enable the program to use the information in the DBRM
that was created in Step 1. This fifth step is the bind process. The bind process
requires DB2, and it uses the DBRM to produce an application plan that enables
the program to access DB2 data. See “The bind process” on page 12 for an
explanation of the bind process.
The operational SQL statements from a DBRM can be placed straight into the plan,
in which case we say that the DBRM is bound into a plan. Alternatively, you can
bind a DBRM into a package (using the BIND PACKAGE command), which
contains the operational SQL statements from a single DBRM. You can group
related packages into collections. You can then include the package name or
collection name in a list of packages, and bind the list of packages into the plan. A
single plan can contain both a package list, and DBRMs bound directly into the
plan. You can create a plan using the DBRMs from a single CICS application, or
you can use the DBRMs from more than one application to create a single plan.
Each thread into DB2 relates to a plan—see “Overview: How threads work” on page
2 for more information about threads. The plan that each type of thread uses is
named on the DB2CONN definition (for pool threads) or the thread’s DB2ENTRY
definition (for entry threads). When CICS requests the use of a thread for an
application to access DB2, it tells DB2 the name of the plan associated with that
type of thread, and DB2 locates the plan. The definition for each type of thread can
either name a specific plan, or it can name a dynamic plan exit, a routine that
determines which plan to use for the transaction that has requested the thread.
If the definition of the pool thread or entry thread names a specific plan, all the
transactions that use that type of thread must use that plan. The transactions that
can use a type of entry thread are specified in the DB2ENTRY and DB2TRAN
definitions for the thread. If the DB2ENTRY definition for the thread names a
specific plan, the DBRMs from all the application programs that could run under all
those transaction IDs must be bound into the same plan, or bound into packages
that are then listed in the same plan. If the DBRMs from any of the application
programs that run under those transaction IDs are bound directly into the plan, and
you change the SQL statements in any of those application programs, the whole
plan will be inaccessible while you bind all the directly-bound DBRMs into the plan
again. This means that no transaction can use that type of entry thread while you
are maintaining the plan. Pool threads could be used by any of your CICS
There are two ways to avoid making types of thread unavailable while you are
maintaining plans. The best solution is to avoid binding DBRMs directly into plans,
by using packages instead. If you bind each separate DBRM as a package and
include them in package lists in plans, the plans are still accessible while you are
maintaining individual packages. While you are carrying out maintenance work on a
particular program, the pool threads or entry threads related to plans involving that
program are still available, because the plans are still accessible. This means that
you can safely name a specific plan for each thread. If you want to start using
packages, see “Using packages” on page 90 and the DB2 Universal Database for
OS/390 and z/OS Application Programming and SQL Guide for details of how to
implement packages.
To use the DB2 JDBC driver shipped with DB2 Versions 7 or 8, the
db2hlq.SDSNLOD2 library also needs to be added to the CICS STEPLIB
concatenation. The DB2 JDBC drivers shipped with DB2 Versions 5 and 6 do not
require this library.
To modify your CICS startup JCL you should concatenate the following libraries on
the STEPLIB DD statement as follows:
v db2hlq.SDSNLOAD (after the CICS libraries)
v db2hlq.SDSNLOD2 (after the CICS libraries)
There should be no DB2 libraries in the DFHRPL DD statement. If DB2 libraries are
required in the DFHRPL concatenation by an application, or by another product,
they should be placed after the CICS libraries.
CICS provides a CICS DB2 attachment facility (the CICS DB2 adaptor) that works
with all supported releases of DB2. The CICS DB2 attachment facility is shipped on
the CICS Transaction Server product tape, and you must use this version of the
attachment facility to connect a CICS Transaction Server region to DB2.
The CICS DB2 attachment facility has been supplied by CICS since CICS/ESA®
4.1. Always use the correct CICS DB2 attachment facility for the release of CICS
under which a region is running—the CICS 4.1 attachment facility for a CICS 4.1
region, and so on.
With the group attach facility, instead of connecting to a specific DB2 subsystem,
you can choose to connect to any one member of a data-sharing group of DB2
subsystems which is active on an MVS image. This allows you to use a common
DB2CONN definition, specifying a group ID, across multiple cloned AORs, and to
reconnect rapidly if the connection to DB2 fails. See “Using the DB2 group attach
facility” on page 51 for more information.
For the CICS DB2 attachment facility to use the open transaction environment
(OTE), CICS must be connected to DB2 Version 6 or later. You must also use
DB2 Version 6 or later “early” (ERLY) code, meaning that the DB2 Version 6 or later
version of db2hlq.SDSNLINK must be present in the MVS link list. If you are using
DB2 Version 6 early code, you must apply APARS PQ43242 and PQ50703. If you
are using DB2 Version 7 early code, you must apply APARS PQ46501 and
PQ50703. (An MVS IPL is required for these DB2 APARS to take effect.)
The open transaction environment enables the CICS DB2 task-related user exit to
execute on an open TCB. Open TCBs, unlike the QR TCB or subtask thread TCBs,
may be used for both non-CICS API requests (including requests to DB2) and
threadsafe application code. Because threadsafe application code can be executed
on the open TCB, a threadsafe CICS DB2 application should not need to switch
between different TCBs several times during the execution of a CICS DB2
application. This situation produces a significant performance improvement where
an application program issues multiple SQL calls. See “Enabling CICS DB2
applications to exploit the open transaction environment (OTE) through threadsafe
programming” on page 106 for more information on the open transaction
environment and its performance benefits.
If you are migrating to DB2 Version 6 or later, note that with effect from DB2
Version 6, the CICS-DB2 language interface module, DSNCLI, is no longer shipped
by the DB2 product. DSNCLI is supplied as part of CICS and is installed in the
SDFHLOAD library when you install CICS TS. Note that DSNCLI is an alias of
DFHD2LI. DSNCLI is also installed in the SDFHAUTH APF-authorized library. In
this library, DSNCLI is an alias of DFHD2LIX.
If you are migrating to DB2 Version 6 or later, and using the open transaction
environment, ensure that the limit set in your MAXOPENTCBS system initialization
parameter is greater than the limit set in the TCBLIMIT attribute of your DB2CONN
definition. In the open transaction environment, MAXOPENTCBS defines the total
number of open TCBs allowed in the CICS system, and TCBLIMIT defines the
number of these open TCBs that can be used to connect to DB2. If your
MAXOPENTCBS limit is lower than your TCBLIMIT, a warning message is issued
when CICS connects to DB2, and you may find that you do not have enough open
TCBs available to process your DB2 workload. In addition, when running with
Transaction Isolation active and connected to DB2 Version 6 or later, set
MAXOPENTCBS to the value of max tasks (MXT) or higher. This will minimise the
possibility of TCB stealing due to a TCB being allocated to the wrong subspace. For
more information, see “The MAXOPENTCBS system initialization parameter and
TCBLIMIT” on page 52.
If you are migrating from a CICS release that defined the CICS DB2 connection
using a resource control table, you now need to define DB2 resource definitions
using CICS resource definition online (RDO). Macro RCTs can still be assembled
for the purpose of migrating them to the DFHCSD file only. The DSNCRCT macro is
shipped with CICS to allow migration of RCT tables to the CSD. It is now wholly
owned by, and incorporated in, CICS . For more information about DSNCRCT, see
CICS Resource Definition Guide.
When describing parameters of the CICS DB2 connection, the names of the
parameters of the RDO objects are now used, not the DSNCRCT macro parameter
names. The CICS Resource Definition Guide describes which DSNCRCT macro
parameter applies to which RDO parameter, and which RDO parameter applies to
which DSNCRCT macro parameter.
All changes made to DB2 resource definitions installed directly from the CSD, or
made by using EXEC CICS CREATE commands, are cataloged and recovered in a
CICS restart. Also, the DB2 objects installed from the CSD remain installed after the
CICS DB2 attachment facility is stopped.
This topic provides information to assist you with the migration process, as follows:
v “If you have not used CICS resource definition online (RDO) before”
v “Effect of migration to RDO on CICS DB2 attachment facility operations” on page
18
v “Effect of migration to RDO on application programs” on page 20
v “Effect of migration to RDO on the INITPARM system initialization parameter” on
page 21
v “Effect of migration to RDO on defaults for resource definition parameters” on
page 21
v “Migrating to RDO for DB2 resource definition” on page 21
If you have not used CICS resource definition online (RDO) before
Using online CICS DB2 resource definition means that you do not have to shut
down the interface between CICS and DB2 when adding, deleting, or changing
CICS DB2 resource definitions. The benefits of using online definition for DB2
resources are discussed in the following sections:
v Function
v System availability
v Performance
Function
The function of the CICS DB2 attachment facility is enhanced by using online
resource definition in the following ways:
System availability
Online CICS DB2 resource definition allows you to add, delete or change definitions
without the need to shut down the interface between CICS and DB2. You are
therefore provided with continuous availability.
Performance
Online CICS DB2 resource definition provides benefits to performance as follows:
v CICS DB2 control blocks are moved above the 16MB line providing virtual
storage constraint relief (VSCR).
v Online CICS DB2 resource definition provides the ability to specify generic
transaction names, using wildcard symbols, which can reduce the number of
definitions required in CICS.
You can use the DSNC STOP <QUIESCE|FORCE> command to stop the CICS
DB2 attachment facility. The QUIESCE option now waits for all active
transactions to complete, that is, new UOWs can start and acquire threads. In
releases of CICS earlier than CICS Transaction Server for OS/390, Version 1
Release 2, a quiesce would only wait for active transactions to release their
thread, which, typically, was at the end of a unit of work (UOW).
During shutdown of the CICS DB2 attachment facility initiated by DSNC STOP,
the terminal remains locked until the stop is complete, when message
DFHDB2025 is issued.
As an alternative to the DSNC command, you can start and stop the CICS DB2
attachment facility using the EXEC CICS SET DB2CONN
CONNECTED|NOTCONNECTED commands. You can also stop the CICS DB2
attachment facility by starting the CICS-supplied transactions CDBQ and CDBF
from an application program, using an EXEC CICS START command. CDBQ
causes a quiesce close and CDBF causes a force close.
CICS DB2 attachment facility command changes
The pool section of the DB2CONN resource definition does not have a TXID
parameter associated with it. To modify the number of threads allowed on the
pool, use reserved name CEPL on the DSNC MODIFY TRANS command. For
example, issue the following command (where n is the new number of threads).
The DSNC DISP TRAN tttt command now displays all threads running for a
particular transid, instead of all the transactions associated with an RCT entry.
In earlier releases, CICS uses the tttt operand to locate an RCT entry and then
displays all the threads for that entry.
When you use the DSNC DISP STAT command, CICS displays statistics for
DSNC commands on a line beginning ’*COMMAND’. Pool thread statistics are
displayed on a line beginning ’*POOL’.
When modifying an RCT entry using the DSNC MODIFY TRANS tttt command,
specify tttt exactly as it was defined in the RCT. If you defined a generic TXID,
you must refer to the generic name when modifying it with a DSNC command.
For example, if you have transactions called TAB and TAC, but they are defined
generically as TA*, you can modify these on a DSNC command only by their
generic name:
You can use the DISMACP system initialization parameter to disable transactions
that abend with an ASRE abend.
To use the CICS DB2 attachment facility, the minimum migration requirements are:
1. Reassemble your existing RCTs with the DSNCRCT macro from
CICSTS23.CICS.SDFHMAC.
DSNCRCT TYPE=ENTRY,TXID=J+++,RDONAME=J
DSNCRCT TYPE=ENTRY,TXID=D*,RDONAME=D
DSNCRCT TYPE=ENTRY,TXID=(ANDY,BILL,CARL),RDONAME=A
You cannot define multiple generic transids referring to the same entry in the
DSNCRCT macro, and these must be defined explicitly in the CSD using the
DEFINE command.
Note that defining transaction IDs using wildcard characters removes the ability
to collect CICS DB2 statistics on a per transaction basis, as statistics are now
collected for each DB2ENTRY, which represents a group of transactions.
v You can edit the DSNCRCT source macros to include a TYPE=GROUP,
GROUP=groupname, to specify the group in which the utility should create the
DB2 objects. All objects are created in the specified group until another
TYPE=GROUP is encountered. If you omit the GROUPNAME parameter,
DFHCSDUP uses a default group name of RCTxx where xx is the 1 or 2
character RCT suffix.
Manual connection
Connection between CICS and DB2 can be established manually by any one of the
following methods:
v Using the DSNC STRT command.
For information on the DSNC STRT command see “DSNC STRT” on page 48.
v Using a CEMT SET DB2CONN CONNECTED command.
For information on the SET DB2CONN CONNECTED command, see CICS
Supplied Transactions.
v Running a user application that issues an EXEC CICS SET DB2CONN
CONNECTED command. For more information about the EXEC CICS SET
DB2CONN CONNECTED command, see the CICS System Programming
Reference.
Manual disconnection
The connection between CICS and DB2 can be stopped or disconnected by using
any one of the following methods:
v Using the DSNC STOP command.
For information on the DSNC STOP command see “DSNC STOP” on page 46.
v Using a CEMT SET DB2CONN NOTCONNECTED command.
v Running the CICS-supplied CDBQ transaction that issues an EXEC CICS SET
DB2CONN NOTCONNECTED command.
The CDBQ transaction runs program DFHD2CM2. The transaction can be run
directly from the terminal or by using the EXEC CICS START command. No
messages are output to the terminal. The CICS DB2 adapter however outputs
messages to transient data as part of its shutdown procedure.
v Running the CICS supplied CDBF transaction that issues an EXEC CICS SET
DB2CONN NOTCONNECTED FORCE command.
The CDBF transaction runs program DFHD2CM3. The transaction can be run
directly from the terminal or EXEC CICS STARTed. No messages are output to
the terminal. The CICS DB2 adapter, however, outputs messages to transient
data as part of its shutdown procedure.
v Running a user application that issues an EXEC CICS SET DB2CONN
NOTCONNECTED command.
| In situations where only DB2 and a single CICS system are involved with the unit of
| work, CICS is always the coordinator. If further parties are involved—for example,
| by means of a LU6.2 communication link—it is possible that the local CICS system
| is not the overall coordinator of the unit of work; for example, a remote CICS
| system might be the coordinator instead. In this situation, it is possible that both
| DB2 and the local CICS system are indoubt about the outcome of the UOW. If a
| failure occurs in this situation, the local CICS system might shunt the unit of work,
| depending on the definition for the transaction. The unit of work is then considered
| to be shunted indoubt.
Indoubt UOWs are normally resolved automatically when the connection between
CICS and DB2 is reestablished. CICS and DB2 exchange information regarding the
indoubt UOWs; that is, CICS informs DB2 whether the UOW was backed out or
To solve this problem, CICS maintains a history of the last DB2 data-sharing group
member to which it connected, which is cataloged and maintained across warm,
emergency and cold starts (but not initial starts). During connection or reconnection
to DB2, the CICS DB2 attachment facility checks this history to see if any
outstanding UOW information is being held for the last DB2 data-sharing group
member to which it connected, and acts as follows:
v If no outstanding UOW information is being held, group attach operates normally
and chooses any active member of the data-sharing group for the connection.
| v If outstanding UOW information is being held, the next action depends on the
| setting you have chosen for the RESYNCMEMBER attribute of the DB2CONN
| definition.
| – If RESYNCMEMBER is set to YES, indicating that you require
| resynchronisation with the last recorded DB2 data-sharing group member,
| CICS ignores the group attach facility and waits until it can reconnect to that
| DB2 data-sharing group member, to resolve the indoubt UOWs. UOWs which
| are shunted indoubt are not included in this process, because CICS itself is
| unable to resolve those UOWs at this time. Resynchronization for those
| UOWs will occur when CICS has resynchronized with its remote coordinator.
| – If RESYNCMEMBER is set to NO, perhaps because you want to reconnect as
| fast as possible, CICS makes one attempt to reconnect to the last recorded
| DB2 data-sharing group member. If this attempt is successful, the indoubt
| UOWs (with the exception of UOWs that are shunted indoubt) can be
| resolved. If it is unsuccessful, CICS uses group attach to connect to any
| active member of the DB2 data-sharing group, and the warning message
| DFHDB2064 is issued stating that there may be unresolved indoubt UOWs
| with the last recorded member.
You should rarely need to initial start CICS. If you simply want to reinstall resources
from the CSD, a cold start should be used, which allows any resynchronization
information to be recovered. In particular, an initial start of CICS should be avoided
if the previous warm shutdown of CICS issued message DFHRM0131 indicating
that resynchronization is outstanding.
If CICS is initial started when DB2 resynchronization was required, when the CICS
DB2 connection is re-established, message DFHDB2001 is output for each UOW
that failed to be resynchronized, and the UOW must be resynchronized in DB2
using the DB2 RECOVER INDOUBT command. CICS Transaction Server has no
equivalent to the DFH$INDB utility that was available in CICS/ESA Version 4 and
earlier, which allowed scanning of the system log to ascertain the outcome of the
UOW. The MVS system log, and hence all the UOW information on it, has been lost
by initial starting of CICS .
The command is routed to DB2 for processing. DB2 checks that the user is
authorized to issue the command entered. Responses are routed back to the
originating CICS user. The command recognition character (CRC) of “-” must be
used to distinguish DB2 commands from CICS DB2 attachment facility commands.
This command recognition character is not used to identify the DB2 subsystem to
which the command is to be sent. The command is sent to the DB2 subsystem to
which CICS is currently connected. Figure 5 shows the CICS DB2 attachment
facility commands. These require CICS authorization to use the DSNC transaction
and the DB2 commands. For more information about the DSNC -DB2COMMAND,
see “Issuing commands to DB2 using DSNC” on page 34.
DISP
DSNC STRT STRT
STOP
MODI
DIS
DSNC DIS REC
THREAD(*) STA
STO
TER
CICS DB2
ADDRESS SPACE ADDRESS SPACE
Figure 5. Examples of CICS DB2 attachment facility commands and some DB2 commands
When CICS is connected to DB2 Version 6 or later, CICS applications that access
DB2 do not enter a CICS wait state, because the CICS DB2 task-related user exit
and the request into DB2 run on an L8 open TCB. In this environment, both purge
and forcepurge of the CICS task are supported. However, with both of these there
is again a risk of terminating a DB2 request during a “must complete” activity in
DB2.
To determine which DB2 thread is associated with a CICS task, use the DSNC
DISPLAY TRAN command (see “DSNC DISPLAY” on page 38), which shows the
CICS task number, transaction id, and the 12–byte DB2 correlation id of the
associated DB2 thread used by the CICS DB2 task. This correlation id uniquely
identifies a thread. A DSNC —DIS THD(*), issues a DB2 display thread command
showing all threads used by this CICS system identified by correlation id, and gives
a unique token that can be used with a DB2 CANCEL THREAD command to cancel
the thread.
DB2 does not produce any reports from the recorded data for accounting,
monitoring, or performance purposes. To produce your own reports you can write
your own programs to process this data, or use the DB2 Performance Monitor
(DB2PM) Program Product.
The DSNC transaction executes program DFHD2CM1, which handles both CICS
DB2 attachment facility and DB2 commands. You can distinguish DB2 commands
from CICS DB2 attachment facility commands by the hyphen (-) character, which is
entered with DB2 commands. This character is not a DB2 subsystem recognition
character, but a command recognition character. It is always a -, independent of the
character actually defining the DB2 subsystem, since CICS can only connect to one
DB2 subsystem at a time. There is no need to use different DB2 subsystem
recognition characters from CICS, and thus we use only the default - character.
DFHD2CM1 can also be activated by transactions other than DSNC. Thus you can
give a different transaction code to each CICS DB2 attachment facility command,
and to each DB2 command. This enables you to apply different levels of security to
each command.Alternative transaction definitions for CICS DB2 attachment facility
commands are supplied in sample group DFH$DB2, using the following names:
DISC
DISP
STRT
STOP
MODI
Alternative transaction definitions for DB2 commands are also supplied in sample
group DFH$DB2, using the following names:
Environment
This command can be issued only from a CICS terminal
Syntax
DSNC syntax
DSNC db2-command
destination
Abbreviation
You can omit DSNC from the DB2 command if the relevant transaction definition is
installed from the CICS DB2 sample group, DFH$DB2. For example, if you are
installing the -DIS transaction definition.
-DIS THD(*)
The sample CICS DB2 group, DFH$DB2, contains the following transaction
definitions for issuing DB2 commands:
Authorization
Issuing DB2 commands using DSNC does not require any authorization from CICS
over and above transaction attach security required to run the DSNC transaction. It
does, however, require DB2 privileges. For more information about CICS security,
see Chapter 6, “Security in a CICS DB2 environment,” on page 65.
Usage note
Screen scrolling
The SIT keywords SKRxxxx can be used to support the scrolling of DSNC DB2
commands from your terminal. For further information about the SIT keywords
and parameters, see the CICS System Definition Guide.
Example
Issue the DB2 command -DISPLAY THREAD from a CICS terminal to display
threads for a CICS with applid IYK4Z2G1.
Environment
This command can be issued only from a CICS terminal.
Syntax
DISC syntax
Abbreviation
DSNC DISC or DISC (using the DISC transaction from the CICS DB2 sample group
DFH$DB2).
Authorization
Access to this command can be controlled using the following CICS authorization
checks:
v Transaction attach security for transaction DSNC
v Command security for resource DB2CONN. This command requires READ
access. For more information about CICS security, see Chapter 6, “Security in a
CICS DB2 environment,” on page 65.
Parameter description
plan-name
Specifies a valid application plan.
Usage notes
Preventing creation of threads
The command DSNC DISCONNECT does not prevent threads from being
created on behalf of transactions. The command only causes currently
connected threads to be terminated as soon as they are not being used by a
transaction. To interrupt a transaction and cancel a thread faster, you can use
the DB2 CANCEL THREAD command.
You can stop the transactions associated with a particular plan ID in CICS with
the MAXACTIVE setting for TRANCLASS. This prevents new instances of the
transaction from causing a re-creation of a thread.
Alternative for protected threads
You may want to deallocate a plan for rebinding or for running a utility against
the database. If you are using a protected thread use EXEC CICS SET
DB2ENTRY(entryname) THREADLIMIT(0), or DSNC MODIFY rather than
Example
Disconnect active and protected threads for plan TESTP05:
Environment
This command can be issued only from a CICS terminal.
Syntax
DISPLAY syntax
Abbreviation
DSNC DISP or DISP (using the DISP transaction from the CICS DB2 sample group
DFH$DB2).
Authorization
Access to this command can be controlled using the following CICS authorization
checks:
v Transaction attach security for transaction DSNC
v Command security for resource DB2CONN. This command requires READ
access. For more information about CICS security, see Chapter 6, “Security in a
CICS DB2 environment,” on page 65.
Parameter description
planname
Displays information about threads by planname. Planname is a valid plan
name for which information is displayed.
If you do not specify planname (or if you specify an asterisk, *), information is
displayed for all active threads.
Example
Display information on all active plan IDs listed in the resource control table.
The display information is to be sent to another terminal designated as
MTO2.
Parameter description
transactionID
A valid transaction ID for which thread information is displayed.
If you do not specify a transaction ID, information is displayed for all active
threads.
Parameter description
Displays the statistical counters associated with each entry in the resource control
table. The counters concern the usage of the available connections of the CICS
DB2 attachment facility to DB2.
Usage notes
If you issue this command from CICS while the CICS DB2 attachment facility is
active but the DB2 subsystem is not, a statistics display is produced with no
obvious indication that the subsystem is not operational. Message DFHDB2037
appears in the CICS message log to indicate that the attachment facility is waiting
for DB2 to start.
Example
Display statistical counters associated with each entry in the resource
control table.
Note that a more detailed set of CICS DB2 statistics can be obtained using
standard CICS statistics interfaces, for example, the commands EXEC
CICS COLLECT STATISTICS and EXEC CICS PERFORM STATISTICS, or
using the DFH0STAT sample program.
Alternative destination
destination
The identifier of another terminal to receive the requested display information. It
must be a valid terminal that is defined to CICS and supported by basic
mapping support (BMS).
Because the optional destination is sometimes preceded by an optional plan
name or transaction ID in the command, each parameter must be unique and
separately identifiable as either a name or a terminal identifier. If only one
parameter is entered, it is first checked to see whether it is a plan name or a
transaction ID, and it is then checked as a destination. To use a character string
that is both a plan name or transaction ID and also a valid terminal identifier,
you must use both the name and destination parameters to display the required
information at the required terminal.
When an alternate destination is specified to receive the requested display
information, the following message is sent to the requesting terminal:
The column named ‘S’ denotes the status of the thread, and can take the following
values:
* The thread is active within a unit of work, and is currently executing in DB2.
A The thread is active within a unit of work, but is not currently executing in DB2.
I The thread is inactive; it is a protected thread that is waiting for new work.
The PLAN associated with the thread is displayed (there is no plan for command
threads).
The PRI-AUTH field shows the primary authorization ID used for the thread. The
SEC-AUTH field shows the secondary authorization ID (if any) for the thread.
The CORRELATION fields shows the 12-byte thread correlation ID which is made
up as eeeettttnnnn where eeee is either COMD, POOL or ENTR indicating whether
it is a command, pool or DB2ENTRY thread; tttt is the transid, and nnnn is a unique
number.
If the thread is active within a unit of work, the CICS transaction name, its task
number and finally the CICS local unit of work ID is displayed.
The correlation ID used in this display is also output on DB2 commands such as
DISPLAY LOCK. For example, by using this display in conjunction with a display
locks command you can find out which CICS task is holding a lock within DB2.
DB2ENTRY
Name of the DB2ENTRY, or “*COMMAND” for DSNC command calls, or
“*POOL” for pool statistics.
PLAN
The plan name associated with this entry. Eight asterisks in this field indicate
that this transaction is using dynamic plan allocation. The command processor
transaction DSNC does not have a plan associated with it.
If a plan name associated with an entry is dynamically changed, the last plan
name is the one put into use.
CALLS
The total number of SQL statements issued by transactions associated with this
entry.
AUTHS
The total number of signon invocations for transactions associated with this
entry. A signon does not indicate whether a new thread is created or an existing
thread is reused. If the thread is reused, a signon occurs only if the
authorization ID or transaction ID has changed.
W/P
The number of times that all available threads for this entry were busy. This
value depends on the value of THREADWAIT for the entry. If THREADWAIT is
set to POOL, W/P indicates the number of times the transaction overflowed to
the pool. An overflow to the pool only shows up in the statistics for the
individual DB2ENTRY, and is not reflected in the pool statistics.
If THREADWAIT is set to YES, this reflects the number of times that either the
transaction had to wait for a thread (because the number of active threads had
reached THREADLIMIT), or the transaction could not use a new thread TCB
(because the number of TCBs in use running threads had reached TCBLIMIT).
HIGH
The maximum number of threads acquired by transactions associated with this
entry at any time since the connection was started, that is, a high watermark of
threads.
Note: In releases of CICS before CICS Transaction Server for OS/390, Version
1 Release 2, the HIGH value also included the transactions forced to
wait for a thread or those diverted to the pool. From CICS Transaction
Server for OS/390, Version 1 Release 2 onwards, the HIGH value only
represents threads actually created on the entry.
Environment
This command can be issued only from a CICS terminal.
Syntax
MODIFY syntax
Abbreviation
DSNC MODI or MODI (using the MODI transaction from the CICS DB2 sample
group DFH$DB2).
Authorization
Access to this command can be controlled using the following CICS authorization
checks:
v Transaction attach security for transaction DSNC.
v Command security for resource DB2CONN. This command requires UPDATE
access.
v For DSNC MODIFY TRANSACTION commands modifying the attributes of a
DB2ENTRY. There are also command security checks for resources DB2ENTRY
and DB2TRAN and resource security for the DB2ENTRY. The command requires
READ access to resource DB2TRAN, and UPDATE access to resource
DB2ENTRY for command security. In addition, the resource security command
requires UPDATE access to the particular DB2ENTRY involved. For more
information about CICS security, see Chapter 6, “Security in a CICS DB2
environment,” on page 65.
Parameter description
DESTination
Specifies that the MSGQUEUE parameter of the DB2CONN table is to be
changed, replacing the ″old″ destination ID with the ″new″ destination ID.
old Any destination ID currently set in the MSGQUEUE of the DB2CONN.
new A new destination identifier.
TRANsaction
Specifies that the THREADLIMIT value associated with the given transaction or
group is to be modified.
transaction-ID
The command uses a transaction ID to identify either the pool, command,
or the DB2ENTRY THREADLIMIT value to be modified.
Usage notes
The integer specified in the command DSNC MODIFY TRANSACTION cannot be
larger than the value specified for the TCBLIMIT parameter of the DB2CONN. The
lowest possible value is zero.
Examples
Example 1
To change the specification of the MSGQUEUE parameter in the
DB2CONN from MTO1 to MTO2 as follows:
DFHDB2039 07/09/98 14:47:17 IYK4Z2G1 The error destinations are: MT02 ****
****.
Example 2
To change the pool thread limit to 12:
Figure 11. Sample output from DSNC MODIFY TRANSACTION command (pool thread)
Example 3
To change the command thread limit to 3:
Figure 12. Sample output from DSNC MODIFY TRANSACTION command (for a command
thread)
Figure 13. Sample output from DSNC MODIFY TRANSACTION command (changing
DB2ENTRY thread limit)
Environment
This command can be issued only from a CICS terminal.
Syntax
STOP syntax
FORCE
DSNC STOP
QUIESCE
Abbreviation
DSNC STOP or STOP (using the STOP transaction from the CICS DB2 sample
group DFH$DB2).
Authorization
Access to this command can be controlled using the following CICS authorization
checks:
v Transaction attach security for transaction DSNC
v Command security for resource DB2CONN. This command requires UPDATE
access. For more information about CICS security, see Chapter 6, “Security in a
CICS DB2 environment,” on page 65.
Parameter description
QUIESCE
Specifies that the CICS DB2 attachment facility is to be stopped after CICS
transactions currently running complete. QUIESCE waits for all active
transactions to complete, so new UOWs can start and acquire threads.
FORCE
Specifies that the CICS DB2 attachment facility is to be stopped immediately by
forcing disconnection with DB2, regardless of any transactions that are running.
Currently running transactions that have accessed DB2 are forcepurged. This
includes transactions that may have committed updates to DB2 in a previous
UOW, but have not yet accessed DB2 in their current UOW.
Usage notes
For a DSNC STOP QUIESCE, message DFHDB2012 is output to the terminal. The
terminal then remains locked until shutdown is complete, when message
DFHDB2025 is output.
For a DSNC STOP FORCE, message DFHDB2022 is output to the terminal. The
terminal then remains locked until shutdown is complete, when message
DFHDB2025 is output.
DSNC STOP
The message resulting from the DSNC STOP command shown in Figure 14
is replaced by the message shown in Figure 15 when shutdown is
complete.
Figure 15. Sample output from DSNC STOP when shutdown is complete
Example 2
To force stop the CICS DB2 attachment facility:
The message resulting from the DSNC STOP FORCE command shown in
Figure 16 is replaced by the message shown in Figure 17 when shutdown
is complete.
Figure 17. Sample output from DSNC STOP FORCE when shutdown is complete
In this example, group attach was used, so the name of the DB2 subsystem
and the name of its group are shown.
Environment
This command can be issued only from a CICS terminal.
Syntax
STRT syntax
DSNC STRT
ssid
Abbreviation
DSNC STRT or STRT (using the STRT transaction from the CICS DB2 sample
group DFH$DB2).
Authorization
Access to this command can be controlled using the following CICS authorization
checks:
v Transaction attach security for transaction DSNC
v Command security for resource DB2CONN. This command requires UPDATE
access. For more information about CICS security, see Chapter 6, “Security in a
CICS DB2 environment,” on page 65.
Parameter description
ssid
Specifies a DB2 subsystem ID to override the DB2 subsystem ID (DB2ID) or
DB2 data-sharing group ID (DB2GROUPID) specified in the DB2CONN. You
cannot specify a DB2 data-sharing group ID in a DSNC STRT command.
Usage notes
If a DB2CONN is not installed when the DSNC STRT command is issued, error
message DFHDB2031 is produced, indicating that no DB2CONN is installed.
Resource definitions must be installed from the CSD before attempting to start the
CICS DB2 attachment facility.
If you issue a DSNC STRT command and specify a DB2 subsystem ID, any
DB2GROUPID in the installed DB2CONN definition is blanked out, and needs to be
set again (using CEDA INSTALL or a SET DB2CONN command) to use group
attach on subsequent occasions.
Examples
Example 1
To start the CICS DB2 attachment facility using the DB2 subsystem ID
(DB2ID) or DB2 data-sharing group ID (DB2GROUPID) from an installed
DB2CONN:
DSNC STRT
In this example, group attach is used, so the name of the DB2 subsystem
and the name of its group are shown.
DFHDB2023I 07/09/98 15:06:07 IYK4Z2G1 The CICS DB2 attachment has connected to
DB2 subsystem DF2D group DFP2
Example 2
To start the CICS DB2 attachment facility, using an installed DB2CONN, but
overriding the DB2 subsystem ID (DB2ID) or DB2 data-sharing group ID
(DB2GROUPID) in the DB2CONN with the DB2 subsystem ID DB3A:
DFHDB2023I 07/09/97 15:06:07 IYK4Z2G1 The CICS DB2 attachment has connected to
DB2 subsystem DB3A
If you are not using group attach, and the DB2 subsystem is not active when an
attempt is made to start the CICS DB2 attachment facility, the following output is
received if STANDBYMODE=NOCONNECT is specified in the DB2CONN:
Figure 20. Sample output from DSNC STRT when DB2 is not active and
STANDBYMODE=NOCONNECT
Figure 21. Sample output from DSNC STRT when DB2 is not active and
STANDBYMODE=CONNECT or RECONNECT
If you are using group attach, and no DB2 subsystems in the data-sharing group
are active when an attempt is made to start the CICS DB2 attachment facility, the
following output is received if STANDBYMODE=NOCONNECT is specified in the
DB2CONN:
DFHDB2037 07/09/01 12:30:10 IYK2ZFV1 DB2 group DFP2 has no active members.
Figure 22. Sample output from DSNC STRT with group attach when no DB2 subsystems in
the data-sharing group are active and STANDBYMODE=NOCONNECT
DFHDB2037 07/09/01 12:55:00 IYK2ZFV1 DB2 group DFP2 has no active members.
The CICS DB2 attachment facility is waiting.
Figure 23. Sample output from DSNC STRT with group attach when no DB2 subsystems in
the data-sharing group are active and STANDBYMODE=CONNECT or RECONNECT
When CICS is connected to DB2 Version 6 or later, and you are exploiting the open
transaction environment, you also need to synchronize your setting for the system
initialization parameter MAXOPENTCBS with the TCBLIMIT value on your
DB2CONN definition. “The MAXOPENTCBS system initialization parameter and
TCBLIMIT” on page 52 has more information about this.
This chapter provides further information to help you decide on your DB2CONN,
DB2ENTRY and DB2TRAN definitions. The topics “What happens during SQL
processing” on page 53 and “How threads are created, used, and terminated” on
page 55 provide more detail about the CICS DB2 environment to help you
understand the effects of your choices. To optimize the performance of your CICS
DB2 connection, “Selecting thread types for optimum performance” on page 60,
“Selecting BIND options for optimum performance” on page 62 and “Coordinating
your DB2CONN, DB2ENTRY, and BIND options” on page 62 provide advice on
coordinating your DB2CONN and DB2ENTRY definitions with the options you
choose during the bind process.
v “Using the DB2 group attach facility”
v “The MAXOPENTCBS system initialization parameter and TCBLIMIT” on page 52
v “What happens during SQL processing” on page 53
v “How threads are created, used, and terminated” on page 55
v “Selecting thread types for optimum performance” on page 60
v “Selecting BIND options for optimum performance” on page 62
v “Coordinating your DB2CONN, DB2ENTRY, and BIND options” on page 62
Group attach is a DB2 facility that allows CICS to connect to any one member of a
data-sharing group of DB2 subsystems, rather than to a specific DB2 subsystem.
The group attach facility chooses any one member of the group that is active on the
local MVS image for the connection to CICS (members that are active on other
MVS images are not eligible for selection). If you use the DB2GROUPID attribute of
the DB2CONN definition to specify the ID for the group of DB2 subsystems, instead
© Copyright IBM Corp. 1997, 2003 51
of using the DB2ID attribute to specify the ID of an individual DB2 subsystem, you
will activate the group attach facility. This means that you can use a common
DB2CONN definition, specifying a group ID, across multiple cloned AORs, and
CICS will connect to any active member of that data-sharing group. See the CICS
Resource Definition Guide for information on how to define and install a DB2CONN
definition.
If you are using group attach and the connection between CICS and DB2 is broken,
CICS might not reconnect to the same DB2 subsystem—it might choose a different
member of the data-sharing group of DB2 subsystems. This means that if indoubt
UOWs are being held by the first DB2 subsystem to which CICS connected, they
cannot be resolved. The RESYNCMEMBER attribute of the DB2CONN definition
can be used to solve this problem. See “Resolving indoubt units of work (UOWs)”
on page 26 for information on the RESYNCMEMBER attribute and how to set it.
After the connection has been established, you can use the CEMT or EXEC CICS
INQUIRE DB2CONN DB2ID() command to find out which member of the
data-sharing group has been chosen for the current connection. See CICS Supplied
Transactions for the CEMT command, and the CICS System Programming
Reference for the EXEC CICS command.
If group attach is set but you want CICS to connect to a specific DB2 subsystem,
you can override group attach. For example, if you want CICS to connect to the
DB2 subsystem with an ID of “xyz”, you can specify the DB2ID using:
v a CEMT or EXEC CICS SET DB2CONN DB2ID(xyz) command (see CICS
Supplied Transactions for the CEMT command, and the CICS System
Programming Reference for the EXEC CICS command)
v a DSNC STRT xyz command (see “DSNC STRT” on page 48)
Each of these methods overrides group attach by setting a DB2ID in the installed
DB2CONN definition.
To ensure that you have enough open TCBs available to meet your DB2 workload,
set the limit in your MAXOPENTCBS system initialization parameter to a value
greater than the limit set in the TCBLIMIT attribute of your DB2CONN definition. If
MAXOPENTCBS is lower than TCBLIMIT, the system may run out of open TCBs
before it reaches TCBLIMIT. When CICS connects to DB2, a warning message,
DFHDB2211, is issued if the CICS DB2 attachment facility detects that CICS is
connecting to DB2 Version 6 or higher, and that the setting of MAXOPENTCBS in
the SIT is lower than the TCBLIMIT setting in the DB2CONN definition. If you
receive this warning message, adjust your MAXOPENTCBS limit.
| In addition, when running with Transaction Isolation active and connected to DB2
| Version 6 or later, set MAXOPENTCBS to the value of max tasks (MXT) or higher.
| This will minimise the possibility of TCB stealing due to a TCB being allocated to
| the wrong subspace. The CICS Performance Guide explains how this happens.
When CICS is connected to DB2 Version 5 or earlier, and is not exploiting the open
transaction environment, the CICS DB2 task-related user exit uses its own specially
created subtask thread TCBs to run threads, rather than using open TCBs. In this
environment, the TCBLIMIT attribute of the DB2CONN definition simply controls
how many subtask thread TCBs the CICS DB2 attachment facility can create.
MAXOPENTCBS is not relevant when CICS is connected to DB2 Version 5 or
earlier, as it only applies to open TCBs. If TCBLIMIT is reached in this environment,
the CICS DB2 task-related user exit must wait until another task stops using a
subtask thread TCB, and it can then use the released subtask thread TCB.
The control block for an application plan, the SKCT, is divided into sections. The
header and directory of an SKCT contain control information; SQL sections contain
SQL statements from the application. A copy of the SKCT, the CT, is made for each
thread executing the plan. Only the header and directory are loaded when the
thread is created, if they are not already in the EDM pool.
The SQL sections of the plan segments are never copied into the CT at thread
creation time. They are copied in, section by section, when the corresponding SQL
statements are executed. For a protected thread with RELEASE(DEALLOCATE),
the CT increases in size until all segments have been copied into the CT.
If the SQL statements for the transaction are bound to a package rather than a
plan, DB2 uses a skeleton package table (SKPT) rather than an SKCT, and a
package table (PT) rather than a CT. The SKPT is allocated when the first SQL
statement is executed; it is not allocated when the thread is created.
SQL processing
The following activities can occur for each SQL statement processed, depending on
thread reuse and the BIND options.
v For the first SQL call in a transaction reusing a thread with a new authorization
ID:
– Signon
– Authorization check
v Load the SKCT SQL section, if it is not already in the EDM pool.
v Create a copy of the SKCT SQL section in the CT, if it is not already there.
v If ACQUIRE(USE) is specified:
– Acquire referenced TS locks, if not already taken.
– Load DBDs in the EDM pool, if they are not already there.
v Process the SQL statement.
If the SQL statement is in a package, the SKPT directory and header are loaded.
The PT is allocated at statement execution time, rather than at thread creation time,
as in the case with the SKCT and the CT for plans bound using
ACQUIRE(ALLOCATE).
Commit processing
The following activities can occur at commit time, depending on your BIND options:
v Release the page locks
Thread release
Transactions release the thread they are using at different times. If the transaction
is terminal-oriented, or non-terminal-oriented and NONTERMREL=YES is specified
in the DB2CONN, the thread is released at SYNCPOINT as well as at end of task
(EOT). This makes it efficient to use a protected thread for transactions issuing
many SYNCPOINTS, if combined with the BIND options ACQUIRE(ALLOCATE)
and RELEASE(DEALLOCATE). In this case the resources to do the following
activities are saved for each syncpoint:
v Terminate and start the thread.
v Release and acquire the TS locks.
v Release and copy segments of the plan into the CT.
Thread termination
The following activities can occur at thread termination time, depending on the
BIND options:
v If RELEASE(DEALLOCATE) is specified:
– Release TS locks
– Free CT pages
v Free work storage.
The CICS DB2 attachment facility uses different types of thread TCB to run threads,
depending on whether CICS is connected to DB2 Version 5 or earlier, or to DB2
Version 6 or later. See “Overview: How threads work” on page 2 for a full
explanation.
When CICS is connected to DB2 Version 5 or earlier, the thread TCB is a subtask
TCB created specially by the CICS DB2 attachment facility to run the thread. This
subtask TCB remains permanently attached to the DB2 connection control block
and thread. In this environment, to reuse the thread, the CICS DB2 attachment
facility must reuse the subtask TCB attached to the thread.
When CICS is connected to DB2 Version 6 or later, the thread TCB is the open
TCB on which the CICS DB2 attachment facility is already running. When it needs
to run a thread, the CICS DB2 attachment facility associates the open TCB with the
DB2 connection control block and thread it wants to use, and the open TCB then
runs the thread. When the thread is no longer needed, the open TCB dissociates
from it, and the DB2 connection control block and thread becomes available for
reuse by another open TCB.
The following general rules apply to thread creation, use, and termination.
When CICS is connected to DB2 Version 6 or later, and is using open TCBs as
the thread TCBs, the rules above do not apply. In this environment, the following
rule applies to the thread TCBs:
v Before an SQL request can be passed to DB2, a thread must be available for the
transaction. The open TCB associates itself with the thread, and becomes the
thread TCB until it dissociates from the thread.
PROTECTNUM(n)
THREADLIMIT(n)
PROTECTNUM(0)
THREADLIMIT(n)
Unprotected entry threads for critical transactions are created, used, and terminated
as follows:
TCB attach
No thread TCBs are attached when the CICS DB2 attachment facility is
started.
A TCB is attached only if needed by a thread. When CICS is connected to
DB2 Version 5 or earlier, the subtask thread TCB normally remains
available for reuse until the CICS DB2 attachment facility is stopped.
Thread creation
A thread is created only when needed by a transaction.
When CICS is connected to DB2 Version 5 or earlier, if no thread is
available, but an unused subtask thread TCB exists for this DB2ENTRY, a
new thread is created and related to the TCB, provided THREADLIMIT is
not exceeded. If no thread is available and no unused TCB is available, a
new TCB is created, and a new thread is built, provided the TCBLIMIT and
THREADLIMIT are not exceeded.
When CICS is connected to DB2 Version 6 or later (and so is using the
open transaction environment), if no thread is available, but an open TCB
can be used as a thread TCB for this DB2ENTRY, a new thread is created
and related to the TCB, provided THREADLIMIT is not exceeded. If
TCBLIMIT is reached, no more open TCBs can be used as thread TCBs for
the DB2ENTRY.
PROTECTNUM(0)
THREADLIMIT(0)
THREADWAIT(POOL)
This is the recommended type of definition for transactions with low volume that do
not require a fast response time. All transactions are forced to use pool threads.
Unprotected entry threads for background transactions are created, used, and
terminated as follows:
TCB attach
No subtask thread TCB is ever attached for this thread definition because
THREADLIMIT=0. A pool thread TCB (or, in the open transaction
environment, an open TCB) is used. All activity related to this entry
definition is forced to a thread and a TCB in the pool. A transaction is then
under the control of the PRIORITY, THREADLIMIT, and THREADWAIT
parameters for the pool. The transaction keeps the PLAN and the
AUTHID/AUTHTYPE values you specified for the entry thread.
Thread creation
A thread is created in the pool when needed by a transaction unless the
THREADLIMIT value for the pool was reached.
Thread termination
The thread is terminated when it is released, unless it has a transaction
queued for it.
Thread reuse
Other transactions using the same plan can reuse the thread, when it
becomes available.
Pool threads
These threads are defined with the following DB2CONN parameter:
THREADLIMIT(n)
As we have read in “How threads are created, used, and terminated” on page 55,
protected entry threads are recommended for:
v High-volume transactions of any type
v Terminal-oriented transactions with many commits
v Non-terminal-oriented transactions with many commits (if NONTERMREL=YES is
specified in the DB2CONN)
Several transactions can be specified to use the same DB2ENTRY. Ideally, they
should all use the same plan. Each low-volume transaction can have its own
DB2ENTRY. In this case thread reuse is not likely and you should specify
PROTECTNUM=0. An alternative is to group a number of low-volume transactions
in the same DB2ENTRY and specify PROTECTNUM=n. This gives better thread
utilization and less overhead.
Authorization checks take place when a thread is created and when a thread is
reused with a new user. Avoiding the overhead in the security check is part of the
performance advantage of using protected threads. This means that from a
performance viewpoint all transactions defined to use the same DB2ENTRY with
PROTECTNUM>0 should use the same authorization ID. At least they should avoid
specifying TERM, OPID, and USERID for the AUTHTYPE parameter, because
these values often vary among instances of a transaction.
It is important that you coordinate your DB2CONN, DB2ENTRY, and BIND options.
For more information, see “Coordinating your DB2CONN, DB2ENTRY, and BIND
options” on page 62.
We recommend that transactions using protected threads should use a plan bound
with ACQUIRE(ALLOCATE) and RELEASE(DEALLOCATE) to reduce the amount of
work done. Although ACQUIRE(ALLOCATE) and RELEASE(DEALLOCATE) reduce
the amount of processing for a protected thread, there are some locking
considerations (for more information, see “Developing a locking strategy in the
CICS DB2 environment” on page 104). The plans for infrequently used transactions
that do not use a protected thread should not typically use ACQUIRE(ALLOCATE),
unless most of the SQL statements in the plan are used in each transaction.
The bind option VALIDATE can also affect performance. You should use
VALIDATE(BIND) in a CICS environment. For more information, see “Bind options
and considerations for programs” on page 138.
It is important that you coordinate your DB2CONN, DB2ENTRY, and BIND options.
For more information, see “Coordinating your DB2CONN, DB2ENTRY, and BIND
options.”
In general it is recommended that you initially set your DB2CONN, DB2ENTRY and
BIND options to the values shown in Table 1. You may find that you get better
performance from other combinations for your own transactions. For each
transaction type a recommended thread type and BIND option are shown. There
are also recommendations for whether transactions should overflow to the pool.
Table 1. Recommended combinations of DB2CONN, DB2ENTRY and BIND options
Transaction Description Thread Type Overflow ACQUIRE RELEASE
High volume (all types) Protected Entry Note 1 ALLOCATE DEALLOCATE
Table 2 shows a summary of the activities involved in processing SQL requests for
the three recommended sets of DB2CONN, DB2ENTRY, and BIND specifications.
An “X” indicates a required activity. The table also demonstrates the performance
advantage of using protected threads without changing the authorization ID.
Table 2. Activities involved in processing SQL requests for different DB2CONN, DB2ENTRY,
and BIND specifications
Activity Protected Threads Unprotected Threads
Create thread: X X X
SIGNON X (1) X X
Authorization Check X (1) X X
Load SKCT Header X X X
Load CT Header X X X
Acquire all TS locks X
Commit:
Terminate Thread: X X X
Release TS locks X X X
Free CT pages X X
Free work storage X X X
Notes:
X. Required activity
1. Only if new authorization ID
2. Only if SQL section is not already in EDM pool
3. Only if SQL section is not already in Cursor Table
| You can also use RACF, or an equivalent external security manager, to protect the
| components that make up CICS and DB2 from unauthorized access. You can apply
| this protection to DB2 databases, logs, bootstrap data sets (BSDSs), and libraries
| outside the scope of DB2, and to CICS data sets and libraries. You can use VSAM
| password protection as a partial replacement for the protection provided by RACF.
| “CICS system resource security” in the CICS RACF Security Guide gives you more
| information about this.
| Note: In this chapter, we refer to RACF as the external security manager used by
| CICS. Except for the explicit RACF examples, the general discussion applies
| equally to any functionally equivalent non-IBM external security manager.
CICS DB2
ADDRESS SPACE ADDRESS SPACE
Admin.
O CICS DB2 SECURITY
SECURITY or RACF
Sign-on Authority
Trans. auth Privileges
ON
User
O DB2 OBJECTS
Plans
Tables
VSAM SECURITY
| This topic describes how to control users’ access to DB2–related resources in the
| CICS region, as follows:
| v “Controlling users’ access to DB2CONN, DB2TRAN, and DB2ENTRY resource
| definitions” tells you how to control access to resource definitions.
| v “Controlling users’ access to DB2-related CICS transactions” on page 73 tells
| you how to control access to transactions.
| Some of the DB2-related resources in the CICS region are subject to further
| security checking by DB2’s security mechanisms, and this topic tells you where this
| applies.
| For wider information about the CICS security topics discussed here, such as
| command and resource security in general, see the CICS RACF Security Guide.
| Because your CICS region only has one DB2CONN definition, you do not need to
| use resource security to protect it; you can control access to the DB2CONN
| definition using command security. Also, DB2TRAN definitions, for the purpose of
| resource security, are treated as extensions of the DB2ENTRY definition to which
| they refer, and are not defined for resource security in their own right. If you give a
| user permission to access a DB2ENTRY definition, you also give them permission
| to access the DB2TRAN definitions that refer to it.1 For resource security, you
| therefore only need to define your DB2ENTRY definitions to RACF.
| When resource security is enabled for a transaction, the external security manager
| checks that the user ID associated with the transaction is authorized to modify the
| resource that is involved. “Resource security” in the CICS RACF Security Guide has
| more information about this process.
1. In the case where a transaction changes the name of the DB2ENTRY with which a DB2TRAN definition is associated, a double
security check is performed, to verify the user’s authority to modify both the old DB2ENTRY to which the definition referred, and
the new DB2ENTRY to which it will refer.
| For example, you can use the PERMIT command to authorize a group of users
| to modify a protected DB2ENTRY, db2ent1 in class XCICSDB2, with UPDATE
| authority, as follows:
| PERMIT db2ent1 CLASS(XCICSDB2) ID(group1) ACCESS(UPDATE)
| When command security is enabled for a transaction, the external security manager
| checks that the user ID associated with the transaction is authorized to use that
| command to modify the type of resource that is involved. “CICS command security”
| in the CICS RACF Security Guide has more information about this process.
| If you have both resource security and command security enabled for a particular
| transaction, RACF performs two security checks against the user ID. For example,
| if a transaction involves the user issuing a DISCARD command against DB2ENTRY
| definition db2ent1, RACF checks:
| Within a transaction, you can query whether a user ID has access to DB2
| resource types by using the EXEC CICS QUERY RESTYPE(SPCOMMAND)
| command, with the RESID parameter specifying DB2CONN, DB2ENTRY, or
| DB2TRAN.
| “Providing authorization IDs to DB2 for the CICS region and for CICS transactions”
| on page 74 explains how to select and change these authorization IDs. To
| summarize, the authorization IDs that CICS provides to DB2 are set by the
| AUTHID, COMAUTHID, AUTHTYPE and COMAUTHTYPE attributes on
| DB2-related resource definitions, and by the SIGNID attribute on the DB2CONN
| definition for the CICS region. To change the authorization IDs, you first need
| authority to modify the DB2CONN and DB2ENTRY definitions, which might be
| protected by command security or resource security. Surrogate security provides an
| extra layer of protection, because it involves CICS acting on DB2’s behalf to check
| that the user modifying the authorization ID, is permitted to act as a surrogate for
| the existing authorization ID that is specified in the resource definition.
| True surrogate security provides security checking when a user attempts to change
| the SIGNID, AUTHID or COMAUTHID attributes on a DB2CONN or DB2ENTRY
| definition, all of which specify an authorization ID that is used when a process signs
| on to DB2. CICS uses the surrogate user facility of RACF to perform this checking.
| A surrogate user is one who has the authority to do work on behalf of another user,
| without knowing that other user’s password. When a user attempts to change one
| of the SIGNID, AUTHID or COMAUTHID attributes, CICS calls RACF to check that
| the user is authorized as a surrogate of the authorization ID that is presently
| specified on the SIGNID, AUTHID or COMAUTHID attribute.
| You can now control which CICS users can use transactions that access DB2. Add
| the appropriate users or groups of users to the access list for the transaction
| profiles, with READ authority. “Defining transaction profiles to RACF” in the CICS
| RACF Security Guide has some recommendations about this.
| For transactions that issue CICS DB2 attachment facility commands and DB2
| commands, bear in mind that:
| CICS DB2 attachment facility commands do not flow to DB2, so they are not
| subject to any further security checking. They are only protected by CICS
| transaction-attach security. However, DB2 commands, and CICS transactions that
| access DB2 to obtain data, are subject to further stages of security checking by
| DB2’s own security mechanisms, as follows:
| v When a transaction signs on to DB2, it must provide valid authorization IDs to
| DB2. The authorization IDs are checked by RACF or an equivalent external
| security manager.
| v Because the transaction is issuing a DB2 command or accessing DB2 data, the
| authorization IDs that it has provided must have permission to perform these
| actions within DB2. In DB2, you can use GRANT statements to give the
| authorization IDs permission to perform actions.
| In addition, the CICS region itself must be authorized to connect to the DB2
| subsytem.
| “Providing authorization IDs to DB2 for the CICS region and for CICS transactions”
| tells you how to authorize the CICS region to connect to the DB2 subsystem, and
| how to provide valid authorization IDs for transactions.
| “Authorizing users to access resources within DB2 (DB2 commands, plans, and
| dynamic SQL)” on page 82 tells you how to grant permissions to the authorization
| IDs that the transactions have provided to DB2.
|
| Providing authorization IDs to DB2 for the CICS region and for CICS
| transactions
| For the purposes of security, DB2 uses the term “process” to represent all forms of
| access to data, either by users interacting directly with DB2, or by users interacting
| with DB2 by way of other programs, including CICS. A process that connects to or
| signs on to DB2 must provide one or more DB2 short identifiers, called
| authorization IDs, that can be used for security checking in the DB2 address space.
| Every process must provide a primary authorization ID, and it can optionally provide
| one or more secondary authorization IDs. DB2 privileges and authority can be
| granted to either primary or secondary authorization IDs. For example, users can
| create a table using their secondary authorization ID. The table is then owned by
| that secondary authorization ID. Any other user that provides DB2 with the same
| CICS has two types of process that need to provide DB2 with authorization IDs:
| v The overall connection between a CICS region and DB2, which is created by the
| CICS DB2 attachment facility. This process has to go through DB2’s connection
| processing to provide DB2 with authorization IDs.
| v CICS transactions that acquire a thread into DB2. These could be, for example, a
| transaction that is retrieving data from a DB2 database, or the DSNC transaction
| that is issuing a DB2 command. For each CICS transaction, the actual process
| that DB2 sees is the thread TCB, which CICS uses to control a transaction’s
| thread into DB2. These processes have to go through DB2’s sign-on processing
| to provide DB2 with authorization IDs.
| During connection processing and sign-on processing, DB2 sets the primary and
| secondary authorization IDs for the process to use in the DB2 address space. By
| default, DB2 uses the authorization IDs that the process has provided. However,
| both connection processing and sign-on processing involve exit routines, and these
| exit routines allow you to influence the setting of the primary and secondary
| authorization IDs. DB2 has a default connection exit routine and a default sign-on
| exit routine. You can replace these with your own exit routines, and a sample
| connection exit routine and sign-on exit routine are supplied with DB2 to assist you
| with this.
| “Providing authorization IDs to DB2 for a CICS region” tells you how to set up
| authorization IDs for a CICS region to connect to DB2.
| “Providing authorization IDs to DB2 for CICS transactions” on page 77 tells you how
| to set up authorization IDs for CICS transactions.
| The user ID that a CICS region might use must be defined to RACF, or your
| equivalent external security manager, if the external security manager is active.
| Define the user ID to RACF as a USER profile. It is not sufficient to define it as a
| RESOURCE profile.
| Once you have defined the CICS region’s user ID to RACF, permit it to access
| DB2, as follows:
| 1. Define a profile for the DB2 subsystem with the single address space
| subsystem (SASS) type of connection, in the RACF class DSNR. For example,
| the following RACF command creates a profile for SASS connections to DB2
| subsystem DB2A in class DSNR:
| RDEFINE DSNR (DB2A.SASS) OWNER(DB2OWNER)
| 2. Permit the user ID for the CICS region to access the DB2 subsystem. For
| example, the following RACF command permits a CICS region with a user ID of
| CICSHA11 to connect to DB2 subsystem DB2A:
| PERMIT DB2A.SASS CLASS(DSNR) ID(CICSHA11) ACCESS(READ)
| DB2’s connection exit routine takes the primary authorization ID (the user ID)
| provided by the CICS region, and sets it as the primary ID for the CICS region in
| DB2. The default DB2 connection exit routine DSN3@ATH, and the sample DB2
| connection exit routine DSN3SATH, both behave in this way. It is possible to
| change the primary ID that DB2 sets, by writing your own connection exit routine.
| The DB2 Universal Database for OS/390 and z/OS Administration Guide has more
| information about the sample connection exit routine and about writing exit routines.
| However, you might find it more straightforward to provide secondary authorization
| IDs for the CICS region, and grant permissions to the CICS region based on these,
| rather than on the primary authorization ID.
| To provide the name of the RACF group, or list of groups, to which a CICS region is
| connected, to DB2 as secondary authorization IDs, complete the following steps:
| A key consideration for choosing the authorization IDs that CICS transactions
| provide to DB2 is the security mechanism that you have chosen for security
| If you are using RACF for some or all of the security checking in your DB2 address
| space, CICS transactions that sign on to DB2 must provide an authorization ID by
| one of the following methods:
| v Specify AUTHTYPE(USERID) or COMAUTHTYPE(USERID) in the appropriate
| definition for the thread (DB2ENTRY or DB2CONN), to provide the user ID of the
| CICS user associated with the transaction to DB2 as the primary authorization
| ID.
| v Specify AUTHTYPE(GROUP) or COMAUTHTYPE(GROUP) in the appropriate
| definition for the thread (DB2ENTRY or DB2CONN), to provide the user ID of the
| CICS user associated with the transaction to DB2 as the primary authorization
| ID, and the name of a RACF group or list of groups as the secondary
| authorization IDs.
| CICS must also be using RACF (SEC=YES must be specified in the SIT). These
| conditions apply because when RACF is used for security checking in the DB2
| address space, CICS needs to pass a RACF access control environment element
| (ACEE) to DB2. CICS can only produce an ACEE if it has RACF active, and only
| threads defined with the USERID or GROUP option can pass the ACEE to DB2.
| Note that if the RACF access control environment element (ACEE) in the CICS
| region is changed in a way that affects the CICS DB2 attachment facility, DB2 is not
| aware of the change until a sign-on occurs. You can use the CEMT or EXEC CICS
| SET DB2CONN SECURITY(REBUILD) command to cause the CICS DB2
| attachment facility to issue a DB2 sign-on the next time a thread is reused, or when
| a thread is built on an already signed-on TCB. This ensures that DB2 is made
| aware of the security change.
| Before you start to set primary authorization IDs, ensure that you have authority to
| do so. As well as having authority to change your DB2CONN or DB2ENTRY
| definitions, if surrogate user checking is in force for the CICS region (that is, the
| system initialization parameter XUSER is set to YES), you need to obtain special
| authority to perform operations involving DB2 authorization IDs. These operations
| are modifying the AUTHID, COMAUTHID, AUTHTYPE, or COMAUTHTYPE
| attributes on a DB2ENTRY or DB2CONN definition, and modifying the SIGNID
| There are two methods of setting the primary authorization ID for a particular type
| of thread:
| 1. Use the AUTHID attribute in the DB2ENTRY definition (for entry threads), or the
| AUTHID or COMAUTHID attribute in the DB2CONN definition (for pool threads
| or command threads), to specify a primary authorization ID. For example, you
| could define AUTHID=test2. In this case, the CICS DB2 attachment facility
| passes the characters TEST2 to DB2 as the primary authorization ID.
| Using AUTHID or COMAUTHID does not permit the use of secondary
| authorization IDs, and also is not compatible with the use of RACF, or an
| equivalent external security manager, for security checking in the DB2 address
| space.
| 2. Use the AUTHTYPE attribute in the DB2ENTRY definition (for entry threads), or
| the AUTHTYPE or COMAUTHTYPE attribute in the DB2CONN definition (for
| pool threads or command threads), to instruct CICS to use an existing ID that is
| relevant to the transaction as the primary authorization ID. This ID can be a
| CICS user ID, operator ID, terminal ID, or transaction ID; or it can be an ID that
| you have specified in the DB2CONN definition for the CICS region.
| Using AUTHTYPE or COMAUTHTYPE is compatible with the use of RACF (or
| an equivalent external security manager) for security checking in the DB2
| address space, if you use the USERID or GROUP options, and with the use of
| secondary authorization IDs, if you use the GROUP option.
| The two methods of determining the primary authorization ID are mutually
| exclusive; you cannot specify both AUTHID and AUTHTYPE, or COMAUTHID and
| COMAUTHTYPE, in the same resource definition.
| Remember that all IDs that you select as primary authorization IDs must be defined
| to RACF, or your equivalent external security manager, if the security manager is
| active for the DB2 subsystem. For RACF, the primary authorization IDs must be
| defined as RACF USER profiles, not just as RESOURCE profiles (for example, as a
| terminal or transaction).
| Follow the instructions in the CICS Resource Definition Guide to set up or modify
| DB2CONN and DB2ENTRY definitions. If you are using the AUTHTYPE or
| COMAUTHTYPE attributes to determine the primary authorization ID for a type of
| thread, use Table 3 on page 80 to identify the options that provide the desired
| authorization ID and support the facilities you want. The key points to consider are:
| v If you want to provide secondary authorization IDs to DB2 as well as a primary
| authorization ID, you need to select the GROUP option. When you specify the
| GROUP option, your primary authorization ID is automatically defined as your
| CICS user ID, but you can base your security checking on the secondary
| authorization IDs instead.
| v If you are using RACF for security checking in the DB2 address space, you need
| to select either the GROUP option, or the USERID option. Only these options
| can pass the RACF access control environment element (ACEE) to DB2, which
| is required when RACF is used for security checking.
| v Think about the performance and maintenance implications of your choice of
| authorization ID. The CICS Performance Guide outlines these. With the USERID,
| OPID, TERM, TX or GROUP options, sign-on processing occurs more frequently,
| and maintenance also takes more time, because you need to grant permissions
| to a greater number of authorization IDs. With the SIGN option, or using the
| Table 3 shows the primary authorization IDs that the CICS DB2 attachment facility
| passes to DB2 when you select each option for the AUTHTYPE or
| COMAUTHTYPE attributes.
| Table 3. Options available on the AUTHTYPE and COMAUTHTYPE attributes
| Option Primary authorization ID passed to Supports Supports
| DB2 RACF secondary
| checking for auth IDs?
| DB2?
| USERID User ID associated with the CICS Yes No
| transaction, as defined to RACF and
| used in CICS sign-on
| OPID User’s CICS operator ID, defined in the No No
| CICS segment of the RACF user profile
| SIGN An ID you specified in the SIGNID No No
| attribute of the DB2CONN definition for
| the CICS region. Defaults to the applid
| of the CICS region
| TERM Terminal ID of the terminal associated No No
| with the transaction
| TX Transaction ID No No
| GROUP User’s CICS RACF user ID used in Yes Yes
| CICS sign-on
|
| If you are not planning to provide secondary authorization IDs for your CICS
| transactions, you do not need to replace the default DB2 sign-on exit routine
| DSN3@SGN. The default sign-on exit routine handles primary authorization IDs.
| However, the DB2 subsystem to which you are connecting might use a different
| sign-on exit routine for some other reason. If the DB2 subsystem uses the sample
| sign-on exit routine DSN3SSGN, you might need to make a change to DSN3SSGN,
| if all of the following conditions are true:
| v You have chosen an AUTHID or AUTHTYPE option other than GROUP.
| v RACF list of groups processing is active.
| v You have transactions whose primary authorization ID is not defined to RACF.
| If this is the case, the DB2 Universal Database for OS/390 and z/OS Administration
| Guide tells you the change you need to make to the sample sign-on exit routine.
| You can only provide secondary authorization IDs to DB2 for CICS transactions if
| you specify the GROUP option for the AUTHTYPE attribute in the DB2ENTRY
| definition (for entry threads), or the AUTHTYPE or COMAUTHTYPE attributes in the
| DB2CONN definition (for pool threads or command threads). If you specify any
| other option for AUTHTYPE or COMAUTHTYPE, the secondary authorization ID is
| set to blanks. When you specify the GROUP option, you cannot choose the primary
| authorization ID for the thread type; it is automatically defined as the user ID of the
| CICS user associated with the transaction. You should base your security checking
| on the secondary authorization IDs instead.
| Access to the DB2 resources that a CICS user needs to perform these actions is
| subject to security checking by DB2’s security mechanisms. In DB2 Version 5 or
| later, you can choose to have this security checking carried out by:
| v DB2’s own internal security.
| v RACF, or an equivalent external security manager.
| v Partly DB2, and partly RACF.
| The DB2 Universal Database for OS/390 and z/OS Administration Guide has more
| information about setting up RACF to perform security checking in the DB2 address
| space.
| If you are using RACF for some or all of the security checking in your DB2 address
| space, remember that CICS transactions that sign on to DB2 must provide an
| authorization ID by using the USERID or GROUP option on the AUTHTYPE or
| COMAUTHTYPE attribute in the resource definition for the thread that they use.
| (“Providing authorization IDs to DB2 for CICS transactions” on page 77 describes
| how to do this.) CICS must also be using RACF (SEC=YES must be specified in
| the SIT). This is because when RACF is used for security checking in the DB2
| address space, CICS needs to pass a RACF access control environment element
| (ACEE) to DB2. CICS can only produce an ACEE if it has RACF active, and only
| threads defined with the USERID or GROUP option can pass the ACEE to DB2.
| When the ACEE is passed to DB2, it is used by the DB2 exit DSNX@XAC, which
| determines whether RACF, or an equivalent non-IBM external security manager, or
| DB2 internal security is used for security checking. DSNX@XAC is driven when a
| transaction whose thread has signed on to DB2, issues API requests. You can
| modify DSNX@XAC — the DB2 Universal Database for OS/390 and z/OS
| Administration Guide has more information about this.
| DB2, or the external security manager, performs security checking using the
| authorization IDs that the CICS transaction provided to DB2 when the thread that it
| was using signed on to DB2. The authorization IDs could be related to the
| individual CICS user (for example, the CICS user’s user ID and the RACF groups to
| which the user is connected), or they could be related to the transaction (for
| DB2, or the external security manager, checks that you have given the authorization
| IDs permission to perform the relevant actions in DB2. You can give the
| authorization IDs this permission by using GRANT statements in DB2. The DB2
| Universal Database for OS/390 and z/OS Administration Guide has full information
| on how to grant, and revoke, DB2 permissions for authorization IDs.
| This topic tells you how to control access to resources in the DB2 address space,
| as follows:
| v “Controlling users’ access to DB2 commands” tells you how to control users’
| ability to issue DB2 commands.
| v “Controlling users’ access to plans” on page 84 tells you how to control users’
| ability to execute plans, and dynamic SQL.
| When a user issues a DB2 command through a CICS transaction, they are also
| subject to DB2’s own security checking, which verifies that they are authorized to
| DB2 to issue the command. This security checking uses the authorization IDs
| (primary or secondary) that the transaction has passed from CICS. “Providing
| authorization IDs to DB2 for the CICS region and for CICS transactions” on page 74
| tells you how to choose these authorization IDs and provide them to DB2. For
| transactions that use DFHD2CM1 to issue DB2 commands, the authorization IDs
| are set by the COMAUTHID or COMAUTHTYPE attribute of the CICS region’s
| DB2CONN definition. For other applications that issue DB2 commands, the
| authorization IDs are set by the AUTHID or AUTHTYPE attribute for the CICS
| region’s resource definition for the type of thread used by the transaction (pool
| thread or entry thread). These attributes control the authorization ID, or type of
| authorization ID, that is passed to DB2 by a transaction that is using that type of
| thread.
| DB2 commands are therefore subject to two security checks, one in the CICS
| address space and one in the DB2 address space. Figure 25 on page 84 illustrates
| the process.
|
Command
CICS INPUT GENERAL
O TERMINAL SECURITY CICS thread COMMAND
CHECKING ATTACH PROCESSOR
CODE
Commands:
DSNC -command DSNC
Authorization
checking
| In most cases, only a limited number of users are permitted to execute DB2
| commands. A convenient solution can be to specify COMAUTHTYPE(USERID) on
| the DB2CONN definition, which resolves to the 8-byte CICS user ID as the
| authorization ID in DB2. Using this method, you can give different DB2 privileges
| explicitly to CICS user IDs. For example, you can use GRANT DISPLAY to give
| specific CICS user IDs permission to use only the -DIS command.
Enter
transaction Authorization
checking for
APPLICATION PROGRAM the plan.
þ EXEC RECEIVE
| However, if a plan includes the use of dynamic SQL, the authorization ID passed
| from CICS to DB2 must possess the privileges required to access all the DB2
| resources involved, both the plan and the data. For example, if you specify
| AUTHTYPE(USERID), the CICS user ID must be granted DB2 privileges to the DB2
| resources involved in the dynamic SQL. If this user ID is also a TSO user ID, it has
| access to the DB2 resources directly from SPUFI, QMF™, and other utilities.
| If you do not want to spend too much time granting DB2 privileges, where a
| transaction executes a plan that involves the use of dynamic SQL, consider using
| one of the following methods of supplying an authorization ID to DB2:
| v Use the SIGN option on the AUTHTYPE attribute of the DB2ENTRY definition for
| the thread used by the transaction. This results in the transaction having the
| primary authorization ID that you specified in the SIGNID attribute of the
| DB2CONN definition for the CICS region. (This method is not suitable where
| RACF is used for security checking in the DB2 address space.)
| v Use the AUTHID attribute of the DB2ENTRY definition for the thread used by the
| transaction, to specify a standard authorization ID. Use the same authorization ID
| for all the transactions that need to access dynamic SQL. (This method is not
| suitable where RACF is used for security checking in the DB2 address space.)
Note that this chapter deals only with the design recommendations that are unique
to the CICS DB2 environment. The general considerations that apply only to CICS
applications, or only to DB2 applications, are not covered. See the DB2 Universal
Database for OS/390 and z/OS Application Programming and SQL Guide for more
information on DB2 application design, and the CICS Application Programming
Guide for more information on CICS application design.
In the design process, decisions can be taken that have consequences related not
only to the application being developed now, but also to future applications. Some
of the key aspects of the design process are as follows:
v Design the relationship between CICS applications and DB2 plans and packages
(see “Designing the relationship between CICS applications and DB2 plans and
packages” on page 88). To gain the greatest benefits for the performance and
administration of your system, the relationship between DB2 plans, transactions,
and application programs must be defined while you are designing applications.
v Develop a locking strategy (see “Developing a locking strategy in the CICS DB2
environment” on page 104). Locking is affected by options you choose when
creating tables in DB2, by the design of your application programs, and by
options you choose when binding plans, so you need to take it into account
throughout the development process.
v Consider the security aspects of both the CICS DB2 test system and the CICS
DB2 production system. The considerations for security in a CICS DB2 system
are described in Chapter 6, “Security in a CICS DB2 environment,” on page 65.
v When developing the application programs, note the considerations in “SQL,
threadsafe and other programming considerations for CICS DB2 applications” on
page 105. The use of certain commands and programming techniques can
improve the performance of your application and avoid some potential problems.
Whether you use qualified or unqualified SQL influences many other aspects of
the CICS DB2 environment. And to gain the performance benefits of the open
transaction environment (OTE), your application programs must be threadsafe.
v If you are using Java™ programs and enterprise beans in the CICS DB2
environment, note the support and programming considerations in Chapter 8,
“Using JDBC and SQLJ to access DB2 data from Java programs and enterprise
beans written for CICS,” on page 117.
v Plan the further steps to be taken when application development is complete and
the application is put into production. Chapter 9, “Preparing CICS DB2 programs
for execution and production,” on page 133 gives information on this process,
including the use of different BIND options.
v Define your CICS DB2 connection to benefit the application’s performance. For
example, using protected entry threads for the application programs to access
DB2 improves performance for heavily-used applications. Chapter 5, “Defining
the CICS DB2 connection,” on page 51 gives information on your CICS DB2
connection.
A well-designed CICS DB2 application should work properly when set into
production for the first time. However, certain factors can affect the application’s
performance later on. These factors include:
v Increased transaction rate
v Continued development of existing applications
v More people involved in developing CICS DB2 applications
v Existing tables used in new applications
v Integration of applications
It is therefore important to develop a consistent set of standards on using DB2 in
the CICS environment.
If you have previously developed applications with data stored in VSAM and
DL/I, be aware that there are several differences between those applications and
CICS DB2 applications. Some of the main differences to consider are:
v Locking mechanism
v Security
v Recovery and restart
v BIND process
v Operational procedures
v Performance
v Programming techniques
One of the major differences between batch and online program design is that
online systems should be designed for a high degree of concurrency. At the same
time, no compromise should be made to data integrity. In addition, most online
systems have design criteria about performance.
To control the characteristics of the plan and the CICS DB2 attachment facility
threads, the relationship between transaction IDs, DB2 plans, and the program
modules must be defined in the design step. Some characteristics of the threads,
environmental description manager (EDM) pool, and plans that depend on the
design are the:
v Plan sizes
There are various design techniques for combining CICS transactions with DB2
plans, as follows:
v The best technique is to use plans based on packages (see “Using packages” on
page 90).
v If you choose not to use packages, you can use one of the following design
techniques for your plans:
– “Using one large plan for all transactions” on page 93
– “Using many small plans” on page 94
– “Using plans based on transaction grouping” on page 95
If you then find that some threads need to be associated with more than one of
your plans (as the pool threads will, unless you have one large plan), you can
use “Dynamic plan exits” on page 96 to achieve this.
v If you need to create the plans for an application after the design and
development process has finished, see “If you need to create plans for an
application that has already been developed” on page 98.
v If a transaction needs to use a program that is not in its plan (perhaps because
the application has changed), see “If you need to switch plans within a
transaction” on page 99.
“A sample application” gives an example of a CICS application accessing DB2, and
this example is used in discussion of the various design techniques.
A sample application
A simple example can be used to explain the consequences of different application
design techniques. Figure 27 on page 90 shows how CICS MAPs and transaction
IDs are correlated, and how the transactions should work, without DB2
considerations.
MAP0 MAPA
RECEIVE RECEIVE
P0 PA
XCTL XCTL
XCTL
P1 P2 P3 PB PC PD
SEND SEND
In this example:
v The transaction ID, TRX0, is specified in the EXEC CICS RETURN
TRANSID(TRX0) command, when a program (not shown) returns control after
displaying MAP0.
v The next transaction then uses the transaction ID, TRX0, independent of what
the terminal user decided to do.
v Program P0 is the initial program for transaction TRX0.
v We assume that all programs shown are issuing SQL calls.
v Depending on the option chosen by the terminal user, program P0 performs a
CICS transfer control (XCTL) to one of the programs: P1, P2, or P3.
v After issuing some SQL calls, these programs display one of the maps: MAP1,
MAP2, or MAP3.
v The example shown on the right side of the figure works in the same way.
v In some situations, program P3 transfers control to program PB.
Using packages
Using packages is the best way to ensure that the relationship between CICS
transactions and DB2 plans is easy to manage. See “Plans, packages and dynamic
plan exits” on page 13 for an overview of plans and packages.
In early releases of DB2, before packages were available, the DBRMs from all the
programs that could run under a particular CICS transaction had to be directly
bound into a single plan. Changing one DBRM in a plan (because the SQL calls for
that program had changed) required all the DBRMs in the plan to be bound again.
Binding a large plan can be very slow, and the entire transaction is unavailable for
Dynamic plan exits (see “Dynamic plan exits” on page 96) were an interim solution
designed to address this problem. The dynamic plan exit is an exit program that
you specify in the DB2CONN or DB2ENTRY definition instead of specifying a plan
name. You create many small plans for your CICS applications, each containing the
DBRMs from a few programs, and the exit program selects the correct plan when
each unit of work begins. You can also use the dynamic plan exit to switch between
plans within a transaction (see “Dynamic plan switching” on page 99). However,
each small plan must still be rebound and taken out of commission every time an
SQL statement changes in one of the programs that uses it. Also, the use of
dynamic plan switching requires an implicit or explicit CICS SYNCPOINT to allow
switching between plans.
Now that packages are available in DB2, using them is the best way to manage
your plans. With packages, you can break the program units into much smaller
parts, which you can rebind individually without affecting the entire plan, or even
affecting the current users of the particular package you are rebinding.
Since updating a plan is easier with packages, you can build much larger
applications without the need to switch transactions, programs, or plans to
accommodate DB2 performance or availability. This also means that you do not
have to maintain as many RDO definitions. You can also avoid situations where you
might otherwise use dynamic SQL to obtain program flexibility. In a plan, you can
even specify packages that do not exist yet, or specify package collections to which
you can add packages. This means that DBRMs from new programs could be
added without disturbing the existing plan at all.
The qualifier option on packages and plans to reference different table sets can
give you more flexibility to avoid plan switching.
In summary, packages:
v Minimize plan outage time, processor time, and catalog table locks
during bind for programs that are logically linked together with START, LINK, or
RETURN TRANSID and have DBRMs bound together to reduce DB2ENTRY
definitions.
v Reduce CICS STARTS or exits.
v Avoid cloning CICS and DB2ENTRY definitions.
v Provide the ability to bind a plan with null packages for later inclusion in the plan.
v Allow you to specify collection at execution time using SET CURRENT
PACKAGESET=variable, which is a powerful feature when used with QUALIFIER
v Provide the QUALIFIER parameter, which adds flexibility to:
– Allow you to reference different table sets using unqualified SQL
– Reduce the need for synonyms and aliases
– Lessen the need for dynamic SQL
There are other benefits that using packages can provide in a CICS environment:
v It allows you to use fewer plans.
v It allows you to bind low-use transactions into a single plan.
v It increases thread reuse potential.
DB2 also provides accounting at the package level. For more information about
packages, refer to the discussions of planning to bind and preparing an application
program to run in the DB2 Universal Database for OS/390 and z/OS Application
Programming and SQL Guide and DB2 Packages: Implementation and Use.
Converting to packages
A transaction that currently uses a dynamic plan exit or dynamic plan switching
techniques can be converted to use packages as follows:
v Bind all of the DBRMs contained in the plan associated with the transaction into
packages in a single collection.
v Bind a new plan with a PKLIST containing a single wildcard entry for this
collection.
v Modify the DB2ENTRY entry for this transaction to use the new plan. Protected
threads can now be used for this transaction to optimize thread reuse.
You could choose to have a single plan for the whole application, or one plan per
transaction. The following sections give more detailed instructions for converting
your transactions, based on this choice.
A similar approach can be taken for converting all CICS applications, whether they
use a dynamic plan exit or not.
Using one plan for the application: This approach gives greatest flexibility in
defining the DB2ENTRYs and DB2TRANs for the application, because the
transactions involved can be grouped to better utilize protected threads and
optimize thread reuse. The steps in converting to this environment are:
1. Bind all DBRMs for the transactions in the application into packages using a
single collection such as COLLAPP1.
2. Bind a new plan, PLANAPP1, with a package list consisting of a single entry,
COLLAPP1.*.
PLANEXITNAME=DSNCUEXT
with
PLAN=PLANAPP1
Using one plan per transaction: This approach was preferable prior to DB2
Version 3 because accounting at the individual package level was not possible,
or
Therefore you should take care to ensure that any modifiable special register is
reset to its initial value before the SYNCPOINT is taken if you want to use dynamic
plan switching. The DB2 Universal Database for OS/390 and z/OS Administration
Guide lists the modifiable special registers.
Disadvantages
v The complete plan must be rebound for any DB2 program modification.
v BIND can be time-consuming for large plans.
v The BIND process cannot take place while the plan is in use. The plan is likely to
be in use in a production system most of the time due to normal activity. In a test
environment, the transaction rate is normally low, but programmers can use
debugging tools that make the response times longer with conversational
programs. This can effectively keep the thread and plan busy.
v DB2-invoked REBIND (due to plan invalidation) allows no DB2 transactions to
execute this plan.
v There is no real information value in messages and statistics pointing out the
plan name, because there is only one plan.
v EDMPOOL must be large enough to cope with DBDs, SKCTs, and CTs and must
allow some fragmentation. Remember that the plan segmentation feature allows
DB2 to load into CTs only parts of the application plans being executed.
Nevertheless, the header and directory parts of an application plan are loaded in
their entirety into the SKCT (if not already loaded), then copied from the SKCT to
CTs. This happens at thread creation time.
Because the application plan directory size is directly dependent on the number
of segments in the plan, using a large plan influences the EDMPOOL size and
the number of I/O operations needed to control it.
We recommend that you use packages rather than using many small plans.
Using many small plans implies either that the program flow follows a narrow path
with limited possibilities for branching out, or that plan switching takes place
frequently.
In the example in Figure 27 on page 90, the switching could take place between
program P0 and the programs at the next lower level, or between program PA and
the programs at the next lower level.
v PLAN1 for (TRX0) using the DBRMs from programs P0, P1, P2, and P3.
v PLANA for (TRXA) using the DBRMs from programs PA, PB, PC, and PD.
However, program P3 can transfer control (using the XCTL command) to program
PB. A plan switching technique must then be used. These techniques are described
in “If you need to switch plans within a transaction” on page 99.
If a DB2 plan is associated with each transaction ID, the application ends up with
many small plans.
Advantages
v Plan maintenance is relatively simple, because little overlap exists between
plans.
v High information value in messages, statistics, and so on, pointing out the plan
name.
Disadvantages
v Plan switching occurs often, unless the application flow follows a narrow path.
v It is difficult to use protected threads, because the transactions are spread over
many sets of transaction IDs, plans, and threads.
v Resource consumption can be high, due to plan switching and low thread reuse.
It is often possible to define such groups of programs, where the programs inside a
group are closely related. That means that they are often executed in the same
transaction, or in different transactions being executed consecutively. One separate
plan should then be used for each group.
In this case, the result of the transaction grouping technique matches the result for
the technique of using many small plans. This is because of the simplicity of the
example used. Normally the transaction grouping technique should produce a larger
plan.
Advantages
v The plan size and the number of different plans can be controlled by the user.
v Thread reuse is likely, depending on the transaction rate within the group.
Disadvantages
v Plan overlap can occur.
v The optimum grouping can change over time.
v Plan switching may be necessary.
Normally, a dynamic plan exit is driven to determine which plan to use at the start
of the first unit of work (UOW) of the transaction. This is referred to as dynamic
plan selection.
A dynamic plan exit can also be driven at the start of a subsequent UOW
(assuming the thread was released at syncpoint) to determine what plan to use for
the next UOW. The plan exit can decide to use a different plan. This is referred to
as dynamic plan switching. See “Dynamic plan switching” on page 99 for more
information.
When a dynamic plan exit is used, DB2 plan allocation occurs only upon execution
of the first SQL statement in a program, or after the program issues a syncpoint and
links or transfers control to another program with a separate DBRM.
IBM supplies two sample assembler language exit programs, DSNCUEXT and
DFHD2PXT, in both source and object code form. You can also write other exit
programs.
The two programs are supplied because as explained in “Enabling CICS DB2
applications to exploit the open transaction environment (OTE) through threadsafe
programming” on page 106, when CICS is connected to DB2 Version 6 or later and
exploits the open transaction environment (OTE), the CICS DB2 task-related user
exit operates as a threadsafe program and is able to receive control on an open
TCB (L8 mode). If the application program that made the DB2 request is
threadsafe, it can also run on the open TCB. In this situation, no TCB switching
should be needed. However, if a dynamic plan exit is used that is defined with
CONCURRENCY(QUASIRENT), like DSNCUEXT, this causes a switch back to the
QR TCB, incurring an additional cost. Dynamic plan exits that are defined with
CONCURRENCY(THREADSAFE) and are coded to threadsafe standards, like
DFHD2PXT, can run on the open TCB, and do not incur the additional cost.
The object code (load module) for the sample plan exits, DSNCUEXT and
DFHD2PXT, are included in the SDFHLOAD library. DSNCUEXT is the default
dynamic plan exit, and it is invoked as a CICS user-replaceable program. The
source code for both sample programs is written in assembler language and
supplied in the SDFHSAMP library. The sample programs show how to address the
parameter list but do not change the plan name.
Before calling the dynamic plan exit, the CICS DB2 attachment facility sets
CPRMPLAN to the name of the DBRM set in the parameter list of the first EXEC
SQL statement executed in the unit of work. As supplied by CICS, the dynamic plan
exits DSNCUEXT and DFHD2PXT do not modify the plan name as input in
CPRMPLAN by the CICS DB2 attachment facility, but return immediately, leaving
the plan name as that chosen by the CICS DB2 attachment facility.
As a consequence of adding support for JDBC and SQLJ support for Java
applications for CICS, the CICS-supplied dynamic plan exits have been changed.
SQLJ and JDBC require that DB2 produce four DBRMs for each application
program in order to support dynamic change of isolation levels. For JDBC and
SQLJ applications, the DBRM name is restricted to seven characters, the eighth
character being used as a suffix of 1,2,3 or 4. Hence for JDBC and SQLJ
applications it is not possible to use a default naming convention of program name
= dbrm name = plan name, as the DBRM name being used will contain a suffix of
1,2,3 or 4.
In order to support a default naming convention for JDBC and SQLJ, the
CICS-supplied dynamic plan exits DSNCUEXT and DFHD2PXT have been changed
to detect an input CPRMPLAN name whose first seven characters are “DSNJDBC”
or “DSNSQLJ”. If such a plan name is detected, the plan name is changed to
“DSNJDBC ”, (with the eighth character set to blanks). Users wishing to use the
default dynamic plan exits with Java applications for CICS should bind the multiple
DBRMs into a plan called DSNJDBC.
The CICS DB2 attachment facility program passes a parameter list to the exit
program using a COMMAREA. The exit program can change the default plan name
(DBRM name) supplied in the parameter list, when the first SQL statement is
processed by the CICS DB2 attachment facility. The name specifies the plan name
for this execution of the transaction.
If you need to create plans for an application that has already been
developed
You can use this technique if the applications were developed with little attention to
the DB2 plan aspects. After the application is completely developed, the plans are
defined to match the transaction.
In general, defining plans after the application has already been developed is not
recommended, but this technique is useful for conversion projects, where the
application design is unchanged but the application now uses DB2.
When defining the DB2 plans and the DB2ENTRY specifications, you can perform
the following steps:
Applied to the example in Figure 27 on page 90, a possible solution would be:
v One plan, PLAN0, using the DBRMs from P0, P1, P2, P3, and PB, used by the
transaction ID TRX0
v One plan, PLANA, using the DBRMs from PA, PB, PC, and PD, used by the
transaction ID TRXA
v Two DB2ENTRY definitions, one for each plan
The best solution to this problem is to use packages. You can bind the DBRM from
the missing program into a package, and add the package to a package list in the
existing plan. The transaction can now access the missing program without needing
to switch plans. There are two other possible solutions, which are discussed below,
as follows:
v “Dynamic plan switching”
v “Switching transaction IDs in order to switch plans” on page 100
In releases of CICS before CICS Transaction Server for OS/390, Version 1 Release
2, dynamic plan switching could only occur for the pool threads, or for RCT entries
that specified THRDA=0, that is, overflowed to the pool. An RCT entry with THRDA
> 0 was not capable of dynamic plan switching. In CICS Transaction Server for
OS/390, Version 1 Release 2 and later versions and releases, dynamic plan
switching can occur for entry threads as well as for the pool, irrespective of the
THREADLIMIT parameter.
If you have coded your own dynamic plan exit, check that the logic copes with
subsequent invocations for the same task. Either the user application or the
dynamic plan exit must be written to tolerate consequences of additional calls to the
exit. If the dynamic plan exit would change the plan when not wanted, the user
application can avoid this by ensuring the thread is not released at syncpoint.
Preferably, if the thread is released, the dynamic plan exit must provide the proper
plan for the new cases when it is called, that is, a DB2ENTRY with THREADLIMIT
> 0.
To invoke the dynamic plan exit to do plan switching after the end of a UOW, your
transaction must release the thread at syncpoint. A transaction releases a thread at
syncpoint only if:
v It is a terminal driven task, or a nonterminal driven task and
NONTERMREL=YES is set in the DB2CONN
v No held cursors are open
v Any DB2 special registers modified have been set back the their initial state.
v DB2 special register CURRENT DEGREE has ever been modified by this
transaction.
If you have to switch transaction IDs, note that in most cases, the first program
transfers data to the next program. The preferred method of doing this is to use an
EXEC CICS RETURN IMMEDIATE command. Alternatively, you can start a new
CICS task against the same terminal, using an EXEC CICS START command, or
using a transient data queue with a trigger level of one. The old program should
issue RETURN to CICS to make the new task start. For both of these switching
techniques, the work done in one user transaction is split up into more than one
UOW. If the new task is backed out, the work done in the first task remains
committed.
If you switch transaction IDs in order to switch plans, the application programs
contain the logic to decide when to switch transaction ID. This means that if you
want to change the plan structure (for example for performance and operational
reasons), you need to change the application programs as well.
Note: It is recommended that you use packages rather than this technique to
control program flow.
Table 5 shows an example of the contents of a control table. The example is based
on the design situations described in Figure 27 on page 90.
Table 5. Control table for sample application
Function name Program Transaction Plan name New TRANS ID
P0 TRX0 PLAN0 *
Sales P1 TRX0 PLAN0
Order P2 TRX0 PLAN0
Pay P3 TRX0 PLAN0
PA TRXA PLANA *
Price PB TRXA PLANA
Order PC TRXA PLANA
Parts PD TRXA PLANA
Function name
The function name field of the table supplements the program field. It works
in exactly the same way. It is used in the cases where the terminal user
enters a function name in a command field, eventually supplied with a key.
The PFCP program can accept either a function name or a program name.
PFCP then uses the function column to search for a valid transaction.
In this way, the logic to interpret the user’s choices is also removed from
the programs and placed in the table, where it is easy to maintain.
Program
The name of the program described in this row.
Transaction
The transaction ID name under which the program in the program column
can be executed.
Plan name
The plan name field is not used. It is shown for illustration purposes only. It
shows the plan name used by the corresponding transaction.
Chapter 7. Application design and development considerations for CICS DB2 101
New TRANS ID
An * in this column of the table means that the corresponding row can be
used when searching for a new transaction ID to start a given program.
When implementing the definitions in CICS, you should consider the following:
v Previously, many different macro RCTs could be used by a CICS system over a
period of time. Therefore, different RCT names could be used at different times.
The same RCT name could be used by different CICS systems.
v With RDO-defined CICS DB2 definitions, the need for multiple RCTs is removed,
because DB2ENTRYs and DB2TRANS can be installed or discarded as required
using the same DB2CONN.
In this case, the control table could contain a DB2ENTRY name instead of the
DB2CONN. At execution time, the PFCP can use the CICS DB2 SPI commands to
inquire whether this DB2ENTRY was installed. In this way using the name of one
DB2ENTRY to identify the set of definitions installed, identifies the “logical RCTs”.
The control table can be implemented in different ways. The most useful solutions
are probably either a DB2 table or a main storage table.
A DB2 table is the simplest to develop and maintain. One or two SELECT calls are
needed for each invocation of PFCP. These SELECTs should return only one row
and indexes can be used. The data and index pages are referenced often and
probably stay in the buffer pool. The response time impact is thus minimal.
A main storage table is faster to access, at least until a certain number of rows in
the table is reached. It is more complicated to maintain.
PGM = PFCP
Terminal
Map COMMAREA
XCTL
TRX0 EXEC
START
COMMAREA
The flow for the application design used in Figure 27 on page 90 is explained
below:
1. The terminal user sends a transaction to CICS. The transaction ID is TRX0.
2. The transaction definition points to program P0.
3. Program P0 receives the map, does some processing, and decides that
program P3 is needed.
4. Instead of transferring control to program P3 (the DBRM for P3 could be part
of another plan), P0 transfers control to the program flow control program
(PFCP in this example). P0 also passes a COMMAREA.
5. PFCP does a table lookup in the control table to see if program P3 is included
in the plan currently in use (PLAN0). This is done by checking if P3 is
specified in the same row as the current transaction ID (TRX0). In the
example, this is the case (line 4 in the table).
6. PFCP then transfers control to program P3. It also passes the COMMAREA it
received from P0 to P3.
7. P3 processes the necessary SQL calls and finishes either by sending a map to
the terminal or by transferring control to PFCP (not shown) to execute another
program.
8. Assuming that this other program is PB, PFCP again checks whether PB is
allowed to run under the current transaction ID, which still is TRX0.
The table shows that PB must not be executed under TRX0. PFCP then
examines the table to find a transaction ID under which program PB can be
executed. In the example, both TRXA and TEMP are valid transaction IDs.
However, TRXA is pointing to program PA in the transaction definition. The
New_TRANS_ID column of the table shows that only the rows with an * can
be used when searching for a new transaction ID to start a given program. In
this case, it is the TEMP transaction.
There are two possibilities for program names in the RDO transaction definition
entry for the TEMP transaction ID:
v The RDO transaction definition can point directly to the new program (PB).
In this case, there must be a transaction ID for each program that could be
Chapter 7. Application design and development considerations for CICS DB2 103
started in this way. Also, to use the COMMAREA, the program being started
must contain logic to find out whether it is being started by START or by
gaining control from an XCTL.
v The RDO transaction definition can point to a common program, here called
the RELAY program. In this case, one or more transaction IDs can be used.
All of them point to the RELAY program in the RDO transaction definition.
The purpose of the RELAY is to transfer control to the appropriate program.
All these programs are then never begun with START and do not need to
handle this situation.
The solution with the RELAY program is shown in Figure 28 on page 103.
9. PFCP starts the transaction TEMP, passing the COMMAREA.
10. The RELAY program is started. It must use an EXEC CICS RETRIEVE
command to retrieve the COMMAREA.
11. From the COMMAREA, RELAY picks up the program name PB.
12. RELAY transfers control to PB, passing the COMMAREA.
13. The plan switch is completed.
Advantages
v This method allows you to implement different types of application design, such
as using one large plan or many small plans.
v The decision of when to switch plans is taken away from the development
process, and is not part of the coding.
v Only the control table needs to be updated when new programs are set into
production. The existing programs do not need to be changed, even if they can
call the new functions.
v The relationship between the transaction IDs, the DB2 plans, and the programs
can be changed without changing the programs. However, the control table must
then be changed.
v Information from the DB2 catalog (SYSPLAN and SYSDBRM) can be used to
build the control table.
v Alternatively, the control table can be used to generate information about the
DBRM structure of the plans.
v The control table contains information that can assist in defining the DB2ENTRYs
and DB2TRANs in CICS, (if the plan name column is available).
v Other functions can be included in a control table structure, for example
information about which transaction ID to use in the TRANSID option of the
EXEC CICS RETURN command.
Disadvantages
The two major disadvantages of this technique are the costs of designing and
developing the solution and the execution time overhead.
Specifying LOCKSIZE(ANY) allows DB2 to decide if lock escalation can take place
for the table space. The DB2 parameter NUMLKTS is the number of concurrent
locks for a table space. If the number of locks exceeds NUMLKTS, lock escalation
takes place. NUMLKTS should then be set to a value so high that lock escalation
does not take place for normal CICS operation.
If a table space lock is achieved and the plan was bound with
RELEASE(DEALLOCATE), the table space is not released at COMMIT time, as
only page locks are released. This can mean that a thread and plan monopolizes
use of the table space.
Using ANY instead of PAGE gives DB2 the option to use lock escalation for
programs that require many page locks before committing. This is typically the case
in batch programs. DB2 also provides the ability to lock at the row level rather than
the page or tablespace level, thus providing better granularity and reducing lock
contention.
You can override DB2 rules for choosing initial lock attributes by using the SQL
statement LOCK TABLE in an application program. However, you should avoid
using the LOCK TABLE statement, unless it is strictly necessary. If the LOCK
TABLE statement is used in an online program, it can prevent the use of
RELEASE(DEALLOCATE) and of protected threads. If you do use a LOCK TABLE
statement, your plan should use the bind option RELEASE(COMMIT).
Chapter 7. Application design and development considerations for CICS DB2 105
Enabling CICS DB2 applications to exploit the open transaction
environment (OTE) through threadsafe programming
The CICS DB2 attachment facility includes a CICS DB2 task-related user exit,
DFHD2EX1, that is invoked when an application program makes an SQL request. It
manages the process of acquiring a thread connection into DB2, and of returning
control to the application program when the DB2 processing is complete.
When CICS is connected to DB2 Version 5 or earlier, the CICS DB2 task-related
user exit operates as a quasi-reentrant task-related user exit program. It runs on the
CICS main TCB (the QR TCB) and uses its own subtask thread TCBs to run
threads, switching to and from the subtask thread TCBs for each DB2 request.
However, when CICS is connected to DB2 Version 6 or later, the CICS DB2
attachment facility exploits the open transaction environment (OTE), to enable the
CICS DB2 task-related user exit to invoke and return from DB2 without switching
TCBs. In the open transaction environment, the CICS DB2 task-related user exit
operates as a threadsafe and open API task-related user exit program—it is
automatically enabled using the OPENAPI option on the ENABLE PROGRAM
command during connection processing. This enables it to receive control on an
open L8 mode TCB. Requests to DB2 are also issued on the L8 TCB, so it acts as
the thread TCB, and no switch to a subtask TCB is needed. For full details of the
CICS DB2 configuration needed to support the open transaction environment, see
“Migrating to a different release of DB2” on page 15.
In the open transaction environment, if the user application program that invoked
the task-related user exit conforms to threadsafe coding conventions and is defined
to CICS as threadsafe, it can also run on the L8 TCB. Before its first SQL request,
the application program runs on the CICS main TCB, the QR TCB. When it makes
an SQL request and invokes the task-related user exit, control passes to the L8
TCB, and DB2 processing is carried out. On return from DB2, if the application
program is threadsafe, it now continues to run on the L8 TCB.
Where the correct conditions are met, the use of open TCBs for CICS DB2
applications decreases usage of the QR TCB, and avoids TCB switching. An ideal
CICS DB2 application program for the open transaction environment is a threadsafe
program, containing only threadsafe EXEC CICS commands, and using only
threadsafe user exit programs. An application like this will move to an L8 TCB when
it makes its first SQL request, and then continue to run on the L8 TCB through any
amount of DB2 requests and application code, requiring no TCB switching. This
situation produces a significant performance improvement where an application
program issues multiple SQL calls. The gains are also significant when using an
enterprise bean, because when enterprise beans make DB2 requests, they require
additional TCB switches to and from the enterprise bean’s own TCB (see “Using
JDBC and SQLJ in enterprise beans: special considerations” on page 130). If the
application program does not issue many SQL calls, the performance benefits might
not be as significant.
If the execution of the program involves any actions that are not threadsafe, CICS
switches back to the QR TCB at that point. Such actions are non-threadsafe CICS
requests issued by the program, the use of non-threadsafe dynamic plan exits, the
use of non-threadsafe task-related user exits, and the involvement of
non-threadsafe global user exits. Switching back and forth between the open TCB
and the QR TCB is detrimental to the application’s performance.
In order to gain the performance benefits of the open transaction environment for
CICS DB2 applications, you must meet the following conditions:
See the CICS Application Programming Guide for information on how to make
application programs and user exit programs threadsafe. By defining a program to
CICS as threadsafe, you are only specifying that the application logic is threadsafe,
not that all the EXEC CICS commands included in the program are threadsafe.
CICS can ensure that EXEC CICS commands are processed safely by switching to
the QR TCB for those commands not yet converted that still rely on
quasi-reentrancy. In order to permit your program to run on an open TCB, CICS
needs you to guarantee that your application logic is threadsafe.
The EXEC CICS commands that are threadsafe, and so do not involve TCB
switching, are indicated in the command syntax diagrams in the CICS Application
Programming Reference and the CICS System Programming Reference with the
statement “This command is threadsafe”, and are listed in Appendix L of the CICS
Application Programming Reference and Appendix D of the CICS System
Programming Reference.
The table below shows what happens when application programs with different
concurrency attributes invoke the CICS DB2 task-related user exit when CICS is
connected to different versions of DB2.
Chapter 7. Application design and development considerations for CICS DB2 107
Table 6. Combinations of application programs and the CICS DB2 task-related user exit
Program’s CICS DB2 task-related Effect
concurrency user exit’s operation
attribute
QUASIRENT or Quasi-reentrant (when Application program and task-related
THREADSAFE connected to DB2 user exit run under the CICS QR TCB.
Version 5 or earlier) The task-related user exit manages its
own TCBs, switching to and from them
for each DB2 request.
QUASIRENT Threadsafe and open Application program runs under the CICS
API (when connected to QR TCB. Task-related user exit runs
DB2 Version 6 or later) under an L8 TCB, and DB2 requests are
executed under the L8 TCB. CICS
switches to and from the CICS QR TCB
and the L8 TCB for each DB2 request.
THREADSAFE Threadsafe and open OTE exploitation. Task-related user exit
API (when connected to runs under an L8 TCB, and DB2
DB2 Version 6 or later) requests are executed under the L8
TCB. The application program also runs
on the L8 TCB when control is returned
to it. No TCB switches are needed until
the task terminates, or if it issues a
non-threadsafe CICS request which
forces a switch back to the QR TCB.
SQL language
The complete SQL language is available to the CICS programmer with only minor
restrictions. For a detailed description on using the SQL language in a CICS
program, see the DB2 Universal Database for OS/390 and z/OS Application
Programming and SQL Guide.
The reason for these recommendations is that the DB2 catalog pages can be
locked, with a lower concurrency level as a consequence. Also the resource
consumption for these types of SQL statements is typically higher than resource
consumption for static DML SQL statements.
Some of the limitations shown in Table 7 can be bypassed if you develop your own
preprocessor to modify the source code before invoking the DB2 precompiler. This
allows you, for example, to change the creator in the SQL statements.
It is recommended that you use qualified SQL for dynamic SQL statements,
because it is easier to administer.
If you use unqualified SQL, you must decide how to supply the CREATOR to fully
identify the tables and views. There are two possibilities:
v You can use synonyms. The synonym must be created by the authorization id
specified in the DB2ENTRY and DB2CONN. Synonyms can only be created by
the authorization ID itself. That means that you must develop a method to create
the synonyms. You can use a TSO ID with the same ID as the authorization ID
specified in the DB2ENTRY or DB2CONN. Another possibility is to design a
Chapter 7. Application design and development considerations for CICS DB2 109
CICS transaction ID (using the same authorization ID) that itself could do the
CREATE SYNONYM statement. However, neither of these methods is advisable.
v If you do not use synonyms, the CREATOR used in the bind process is the
authorization ID of the binder. All tables and views referenced in the dynamic
SQL must then be created with this ID. All transactions using dynamic SQL to
access a common set of DB2 resources must then have the same authorization
ID specified in the DB2ENTRY or DB2CONN. In most cases, it must be the
SIGNID, or a character string. This restriction is normally not acceptable.
For these reasons, the use of unqualified SQL in dynamic SQL statements is not
recommended.
Views
It is generally recommended that you use views where appropriate. Some views,
however, cannot be updated.
In a real-time, online system, you often need to update rows you have retrieved
using views. If the view update restriction forces you to update the base table
directly (or by using another view), you should consider only views that can be
updated. In most cases this makes the program easier to read and modify.
Commit processing
CICS ignores any EXEC SQL COMMIT statement in your application programs. The
DB2 commit must be synchronized with CICS, which means that your program
must issue an EXEC CICS SYNCPOINT command. CICS then performs the commit
processing with DB2. An implicit SYNCPOINT is always invoked by the EXEC CICS
RETURN at EOT.
Serializing transactions
You may need to serialize the execution of one or more transactions. This typically
occurs when the application logic was not designed to deal with concurrency and in
cases where the risk of deadlocks is too high.
You should allow serialization only for low-volume transactions because of potential
queueing time.
The following methods each have different serialization start and end times:
v CICS transaction classes. The CICS facility of letting only one transaction
execute at a time in a CLASS is useful to serialize the complete transaction.
v DB2 thread serialization. In cases where the serialization may be limited to an
interval from the first SQL call to syncpoint (for terminal-oriented transactions,
and nonterminal-oriented transactions if NONTERMREL=YES is defined), you
can use your DB2ENTRY specifications to ensure that only one thread of a
specific type is created at one time. This technique allows concurrency for the
first part of the transaction, and is useful if the first SQL call is not in the
beginning of the transaction. Do not use this technique if your transaction
updated other resources before it issues its first SQL statement.
v CICS enqueue and dequeue. If you know that the serialization period necessary
is only a small part of the programs, then the CICS enqueue and dequeue
technique can be useful. The advantage is that only the critical part of the
transaction is serialized. This part can be as small as just one SQL statement. It
allows a higher transaction rate than the other methods, because the serialization
is kept to a minimum.
The disadvantage compared to the other techniques is that the serialization is
done in the application code and requires the programs to be changed.
v LOCK TABLE statement. It is recommended that you do not use the LOCK
TABLE statement.
The LOCK TABLE statement can be used to serialize CICS transactions and
other programs, if EXCLUSIVE mode is specified. Note that it is the whole table
space that is locked, not the table referenced in the statement.
The serialization starts when the LOCK statement is executed. The end time for
the serialization is when the table space lock is released. This can be at
syncpoint or at thread deallocation time.
Use this technique with care, because of the risk of locking the table space until
thread deallocation time. However, this technique is the only one that works
across the complete DB2 system. The other techniques are limited to controlling
serialization of only CICS transactions.
Chapter 7. Application design and development considerations for CICS DB2 111
Page contention
When designing applications and databases, consider the impact of having many
transactions accessing the same part of a table space. The term “hot spot” is often
used to describe a small part of the table space, where the access density is
significantly higher than the access density for the rest of the table space.
If the pages are used for SELECT processing only, there is no concurrency
problem. The pages are likely to stay in the buffer pool, so little I/O activity takes
place. However, if the pages are updated frequently, you may find that you have
concurrency problems, because the pages are locked from first update until
syncpoint. Other transactions using the same pages have to wait. Deadlocks and
timeouts often occur in connection with hot spots.
Two examples of hot spots are sequential number allocation and insert in
sequence.
Insert in sequence
In situations where many transactions are inserting rows in the same table space,
you should consider the sequence of the inserted rows. If you base a clustering
index on a field with a time stamp, or a sequential number, DB2 tries to insert all
rows adjacent to each other. The pages where the rows are inserted can then be
considered a hot spot.
Note that in the clustering index, all inserts are also in the same page, within a
given period.
A solution to the deadlock risk is to spread the rows by choosing another index as
clustering.
All locks are released, except for those required to maintain the cursor’s position.
Any exclusive page locks are downgraded to shared locks.
In summary:
v The next FETCH following a syncpoint must come from the same task.
v You cannot hold a cursor across end of task.
v Therefore, cursors are not held across the EOT portions of pseudoconversational
transactions.
If you try to hold a cursor across EOT, the cursor is closed and you get an
SQLCODE -501 when you execute the next FETCH. The precompiler cannot detect
this and you do not get a warning message notifying you of this situation.
In general, threads can become candidates for reuse at each syncpoint. When you
use DECLARE CURSOR...WITH HOLD in the CICS applications, consider the
following recommendations:
Chapter 7. Application design and development considerations for CICS DB2 113
v Close held cursors as soon as they are no longer needed. Once all held cursors
are closed, syncpoint can free the thread for thread reuse.
v Always close held cursors before EOT. If you do not close your held cursors, the
CICS DB2 attachment facility forces signon to restore the thread to the initial
state, and this incurs additional processor time.
The new transaction behaves as if it were started by input from the terminal. You
can pass data to the transaction designated by the TRANSID option, using a
COMMAREA. If you choose to, the transaction issuing the RETURN command can
also pass a terminal input message using the INPUTMSG and INPUTMSGLEN
options. This facility allows you to immediately initiate a transaction that expects to
be initiated as a result of terminal input.
This facility provides the same general capability as that achieved by issuing an
EXEC CICS START TRANSID(....) with TERMID(....) set to the EIBTRMID value,
but with much less overhead and without the momentary keyboard unlocking. The
EXEC CICS RETURN TRANSID() IMMEDIATE command permits a
pseudoconversational transaction to switch transaction codes. This could be
advisable, for example, to keep DB2 plan sizes smaller or to have better accounting
statistics for charge-back purposes.
When the CICS DB2 attachment facility is enabled it is not necessarily connected to
DB2. It can be waiting for DB2 to initialize. When this occurs, and an application
issues an EXEC SQL command when CONNECTERROR=ABEND is specified in
the DB2CONN, an AEY9 abend would result. CONNECTERROR=SQLCODE would
result in a -923 SQL code being returned to the application.
You can use the INQUIRE EXITPROGRAM command with the CONNECTST
keyword in place of the EXTRACT EXIT command to determine whether the CICS
is connected to DB2.
CSTAT DS F
ENTNAME DS CL8
EXITPROG DS CL8
...
MVC ENTNAME,=CL8’DSNCSQL’
MVC EXITPROG,=CL8’DFHD2EX1’
EXEC CICS INQUIRE EXITPROGRAM(EXITPROG) X
ENTRYNAME(ENTNAME) CONNECTST(CSTAT) NOHANDLE
CLC EIBRESP,DFHRESP(NORMAL)
BNE NOTREADY
CLC CSTAT,DFHVALUE(CONNECTED)
BNE NOTREADY
If an application avoids making DB2 calls because it knows the CICS DB2
connection is not active, but issues an error message instead and returns normally,
it could delude the workload manager into routing more work to the CICS region.
This is called the “storm drain effect”. Because the application did not abend, the
workload manager believes that good response times are being achieved by this
CICS region for DB2 work, and routes more work down the “storm drain”.
Chapter 7. Application design and development considerations for CICS DB2 115
v Use CONNECTERROR=SQLCODE provided applications handle the -923 code
correctly.
v Avoid using EXTRACT EXIT or INQUIRE EXITPROGRAM commands if
CONNECTERROR=SQLCODE can be used.
v Use CONNECTERROR=ABEND if an AEY9 abend is required. Use the INQUIRE
EXITPROGRAM command instead of the EXTRACT EXIT command.
v It is worth noting that AEY9 abends can still occur even when
STANDBYMODE=RECONNECT and CONNECTERROR=SQLCODE are
specified if:
– The CICS DB2 attachment facility is never started. An AEY9 results if an
application issues an EXEC SQL command. You should always specify
DB2CONN=YES in the SIT, or program DFHD2CM0 in PLTPI. Therefore the
CICS DB2 attachment is at minimum in standby mode.
– The CICS DB2 attachment is shut down using a DSNC STOP or CEMT/EXEC
CICS SET DB2CONN NOTCONNECTED command.
It is advisable to avoid shutting down the attachment. The CICS DB2 SPI
commands allow dynamic modification of the environment without shutting down the
attachment.
DB2 Version 5 and DB2 Version 6 provide the JDBC 1.2 driver, that supports the
JDBC 1.2 application programming interface. DB2 Version 7 and DB2 Version 8
provide two levels of the JDBC driver, the 1.2 level and the 2.0 level. The JDBC 1.2
The JDBC 2.0 driver supports a selected subset of the JDBC 2.0 application
programming interface, and it is downward compatible, so it supports the JDBC 1.2
API as well. Existing Java applications and enterprise beans that were written using
the JDBC 1.2 API and that run in a JVM, can run using the JDBC 2.0 driver, and
can benefit from performance improvements made in the JDBC 2.0 driver. If you
create new Java applications and enterprise beans that use the JDBC 2.0 API, you
need to use the JDBC 2.0 driver.
The following table summarizes which level of the JDBC driver is supported by
which versions and releases of CICS and DB2.
Table 8. Support for the JDBC driver in CICS and DB2
Product JDBC driver supported
DB2 Version 5 (with APAR PQ56653) 1.2 level
DB2 Version 6 (with APAR PQ56654) 1.2 level
| DB2 Version 7 (with APAR PQ56655) 1.2 level and 2.0 level
DB2 Version 8 1.2 level and 2.0 level
CICS TS for OS/390 Version 1 Release 3 (with APAR 1.2 level
PQ34321)
CICS TS for z/OS Version 2 Release 1 1.2 level
CICS TS for z/OS Version 2 Release 2 (with APAR 1.2 level and 2.0 level
PQ57455)
CICS TS for z/OS Version 2 Release 3 1.2 level and 2.0 level
Note: The JDBC 1.2 level driver supports both Java program objects, and Java programs
and enterprise beans that run in a JVM. The JDBC 2.0 level driver only supports Java
programs and enterprise beans that run in a JVM.
Full details of how to code and build Java applications that use the JDBC and SQLJ
application programming interfaces can be found in the DB2 Universal Database for
OS/390 and z/OS Application Programming Guide and Reference for Java that
applies to your version of DB2. The document number for DB2 Version 6 is
SC26-9018, and the document number for DB2 Version 7 is SC26-9932. Particular
programming features apply to JDBC and SQLJ when they are used in a CICS
environment, so read “Programming with JDBC and SQLJ in the CICS DB2
environment” on page 122 for more specific guidance before developing your Java
application.
Chapter 8. Using JDBC and SQLJ to access DB2 data from Java programs and enterprise beans written for CICS 119
you have any applications that use the supported subset of the JDBC 2.0 API,
you must use the JDBC 2.0 driver. The JDBC 2.0 driver also supports
applications written using the JDBC 1.2 API.
v For both the JDBC 1.2 driver and the JDBC 2.0 driver, you also need to add the
DB2 directory containing the serialized profile DSNJDBC_JDBCProfile.ser either to
the JVM’s standard class path (the CLASSPATH option in the JVM profile), or to
its shareable application class path (the
ibm.jvm.shareable.application.class.path option in the JVM properties file). If
the JVM is a worker JVM that uses the shared class cache, its shareable
application class path is taken from the JVM profile for the master JVM that
initializes the shared class cache, but its standard class path is taken from the
JVM profile for the worker JVM itself. The serialized profile is initially created in
the DB2 directory where the command db2gen jdbc was issued, but it might have
been subsequently moved to another location. The directory that needs to be
added to the JVM’s class path is the directory that currently contains the
serialized profile. When you have added the directory to the JVM’s class path,
you also need to ensure that the DB2 system property
db2sqljjdbcprogram=dsnjdbc is set in the JVM properties file for the JVM. This
system property is the default, so it does not need to be added to the JVM
properties file if it is not present, but it should not be specified with a different
program name.
Requirements for the DB2-supplied JDBC drivers (for all types of Java
programs)
v To use the JDBC 1.2 driver or the JDBC 2.0 driver shipped with DB2 Version 7
or later, you need to add the SDSNLOD2 library to the CICS STEPLIB
concatenation. The JDBC 1.2 drivers shipped with DB2 Versions 5 and 6 do not
require this library.
v To enable your applications to use the DB2-supplied JDBC drivers, you need to
name the drivers by using the jdbc.drivers system property in the JVM
properties files that are referenced by the JVM profiles used by your applications.
The CICS-supplied sample JVM properties files contain commented-out
examples of how to do this. Java Applications in CICS tells you how to locate
and customize JVM properties files. Naming the JDBC drivers removes the need
for applications to load the drivers themselves using the Class.forName()
method. Instead, the DriverManager class loads the required class for the
application. The name of the driver is the same for the JDBC 1.2 driver and for
the JDBC 2.0 driver, so if you follow this method, you do not need to change
your existing applications or your JVM properties files when you move to the
JDBC 2.0 driver.
v When running the JDBC driver in a CICS environment, you might want to alter
the system properties in your JVM properties files to tailor your environment.
With CICS, the DB2 environment variable DB2SQLJPROPERTIES, which names
a system properties file to be used by the JDBC driver, is not used. Instead, you
can set system properties related to the JDBC driver in the JVM properties files.
These are the files named by the JVMPROPS parameter in your JVM profiles. If
you want to alter the system properties, note that most DB2 JDBC driver system
properties are not used in a CICS environment. See Appendix B of the DB2
Universal Database for OS/390 and z/OS Application Programming Guide and
Reference for Java which is appropriate for your version of DB2. The document
number for DB2 Version 6 is SC26-9018, and the document number for DB2
Version 7 is SC26-9932. Appendix B of the document has a list of the properties
that are not used or have a different meaning in a CICS environment. The full list
of DB2 JDBC driver properties is in Chapter 6 of the Version 7 Guide, or Chapter
7 of the Version 6 Guide.
You also need to name your Java 2 security policy file in the JVM properties file,
using the following statement:
java.security.policy= /directory/tree/file.name
CICS supplies an example Java 2 security policy for use with Java programs or
enterprise beans in the CICS environment, and you can find the example file at
/usr/lpp/cicsts/lib/security/dfjejbpl.policy.
v To use JDBC and SQLJ, you need to amend your Java 2 security policy to grant
permissions to the JDBC driver, by adding the following lines:
grant codeBase "file:/usr/lpp/db2710/-" {
permission java.security.AllPermission;
};
where db2710 is the high-level qualifier for your DB2 libraries. The example Java
2 security policy dfjejbpl.policy does not contain this statement, so you need
to add it.
v In your Java 2 security policy, you also need to grant read permissions, by
adding the following lines:
grant {
};
Chapter 8. Using JDBC and SQLJ to access DB2 data from Java programs and enterprise beans written for CICS 121
Programming with JDBC and SQLJ in the CICS DB2 environment
The JDBC 1.2 driver supports the JDBC 1.2 and the SQLJ Part 0 (ANSI 98) level of
application programming interfaces. The JDBC 2.0 driver supports the JDBC 1.2
API and a subset of the JDBC 2.0 API, and the level of support for SQLJ is the
same as for the JDBC 1.2 driver. Java programs for CICS must adhere to the
programming rules of these application programming interfaces, which are more
restrictive than the general CICS programming model.
You can find more information about the JDBC APIs at the JDBC Web site,
http://java.sun.com/products/jdbc. For information about the features of the JDBC
APIs that are supported by the DB2-supplied JDBC drivers, and how to code and
build Java applications that use the JDBC and SQLJ APIs, see the DB2 Universal
Database for OS/390 and z/OS Application Programming Guide and Reference for
Java which is appropriate for your version of DB2. The document number for DB2
Version 6 is SC26-9018, and the document number for DB2 Version 7 is
SC26-9932.
The particular programming features that apply to JDBC and SQLJ when used in a
CICS environment are described in the following sections:
v “Acquiring a connection to a database”
– “Acquiring a connection using the JDBC DriverManager interface” on page
123
– “Acquiring a connection using the DataSource interface” on page 124
v “Committing a unit of work” on page 129
v “CICS abends during JDBC or SQLJ requests” on page 130
v “Using JDBC and SQLJ in enterprise beans: special considerations” on page 130
jdbc:db2os390sqlj:
or
jdbc:default:connection
Typically, the location name is the name of the local DB2 to which CICS is
connected. However, you can specify the name of a remote DB2 to access.
In this case, CICS uses the local DB2 as a pass-through, and uses DB2
Distributed Data Facilities to access the remote DB2.
It is recommended that you use a default URL in a CICS environment. The use of
an explicit URL causes particular behaviours at the close of a connection, that could
be inconvenient when multiple programs are used in the same application suite.
Also, when a default URL is used, the connection behaves in the same way using
either the JDBC 1.2 driver, or the JDBC 2.0 driver. See “Committing a unit of work”
on page 129 for further information.
An application that has an open connection should close the connection before
linking to another application that wants to use JDBC or SQLJ. For Java programs
that are part of an application suite, you need to consider the implications of closing
the connection, because if you are using an explicit URL, closing the connection
can cause a syncpoint to be taken. If you are using a default URL, a syncpoint
does not have to be taken when the connection is closed. See “Committing a unit of
work” on page 129 for more information about this.
To use this method of acquiring a connection, your Java application needs to invoke
the DriverManager.getConnection method to specify a default URL or an explicit
URL, to connect to a DB2 subsystem. (For more information about default and
explicit URLs, see “Acquiring a connection to a database” on page 122.) In a CICS
Chapter 8. Using JDBC and SQLJ to access DB2 data from Java programs and enterprise beans written for CICS 123
DB2 environment, you do not need to specify a userid and password on the
DriverManager.getConnection request. If you do specify these, they are ignored for
DB2 Version 6, but cause an error for DB2 Version 7. The existing CICS DB2
security procedures are used instead.
The DataSource in the sample applications does not specify the database name, so
it generates connections with default URL characteristics. A DataSource that
specified the database name would generate connections with an explicit URL. It is
recommended that you do not specify the database name in your DataSource, so
that it generates JDBC connections with a default URL. Using a default URL
ensures the correct behaviour at the close of a connection. See “Committing a unit
of work” on page 129 for further information.
Chapter 8. Using JDBC and SQLJ to access DB2 data from Java programs and enterprise beans written for CICS 125
2. Use CEDA to install transactions DSDB, DSPU, and DSRE from group
DFH$DB2.
3. Use CEDA to install programs DFJ$DSDB, DFJ$DSPU, and DFJ$DSRE from
group DFH$DB2.
| 4. The default name is jdbc/CICSDB2DataSource. You can change the name of
| the DataSource, or the subContext to where the DataSource will be published.
| There are two ways to do this:
| a. You can change the sample programs CICSDataSourcePublish.java,
| CICSDataSourceRetract.java, and CICSjdbcDataSource.java as required. If
| you change the name or subContext for the DataSource, remember to make
| the same change in all three of the sample programs. Put the new classes
| for the sample programs on your CLASSPATH option in the JVM profile
| DFHJVMPR.
| b. Alternatively, you can add the system property
| com.ibm.cics.datasource.path to the JVM properties file that is referenced
| by the JVMPROPS statement in the JVM profile DFHJVMPR. When you
| install CICS, the CICS-supplied sample JVM properties files are placed in
| the directory /usr/lpp/cicsts/cicsts23/props/, where cicsts23 is the
| value that you chose for the CICS_DIRECTORY variable used by the
| DFHIJVMJ job during CICS installation. The CICS-supplied sample profile
| DFHJVMPR references the JVM properties file dfjjvmpr.props. Include the
| following statement in the JVM properties file:
| com.ibm.cics.datasource.path=subContext/DataSourcename
| Specify the subContext and name that you want to use for the DataSource.
| If the com.ibm.cics.datasource.path system property is present in the JVM
| properties file, the sample applications use the values specified by that property.
| Otherwise, they default to the DataSource name specified in the sample
| programs.
5. Before you use the sample applications, ensure that your user ID is authorized
to access DB2 resources. “Authorizing users to access resources within DB2
(DB2 commands, plans, and dynamic SQL)” on page 82 tells you how to ensure
that your user ID has the correct authorization.
| Although the sample applications use the CICS-supplied sample JVM profile
| DFHJVMPR, which is a profile for a standalone JVM (that does not use the shared
| class cache), DataSource lookup classes based on the sample application
| CICSjdbcDataSource.java can also be used by worker JVMs (that do use the
| shared class cache). Bear in mind that for a worker JVM, the CLASSPATH option in
| the JVM profile and the com.ibm.cics.datasource.path system property in the JVM
| properties file are taken from the JVM profile and JVM properties file for the worker
| JVM itself. However, the trusted middleware class path (including the TMSUFFIX
| statement or the TMPREFIX statement, where you need to specify the
| db2j2classes.zip file) is taken from the JVM profile for the master JVM.
| If the DataSource has already been published, the following message is displayed
| on the screen:
| DSPU - CICSDataSourcePublish: Datasource jdbc/CICSDB2DataSource already published
*********************************************************************************
**** CICSDataSourcePublish: started
**** CICSDataSourcePublish: Looking up DataSource jdbc/CICSDB2DataSource
**** CICSDataSourcePublish: DataSource jdbc/CICSDB2DataSource not found
**** CICSDataSourcePublish: Binding DataSource jdbc/CICSDB2DataSource
**** CICSDataSourcePublish: DataSource bound to JNDI
**** CICSDataSourcePublish: ended
*********************************************************************************
Figure 30. Stdout output from transaction DSPU to publish a DataSource with default name and subContext
The sample program publishes a DataSource that provides the information needed
to generate default URL connections. Now, instead of specifying a URL in a
DriverManager.getConnection request (the JDBC DriverManager interface), your
application programs can use the Java Naming and Directory Interface (JNDI) to
look up a reference to the DataSource, as described in “Looking up a DataSource
using CICSjdbcDataSource.java.” Remember that CICS must be using the JDBC
2.0 driver (provided by DB2 Version 7 or later) to use the DataSource interface.
Chapter 8. Using JDBC and SQLJ to access DB2 data from Java programs and enterprise beans written for CICS 127
*********************************************************************************
**** CICSjdbcDataSource: started
**** CICSjdbcDataSource: Looking up CICS datasource jdbc/CICSDB2DataSource
**** CICSjdbcDataSource: DataSource Connection created.
**** CICSjdbcDataSource: AutoCommit is false
**** CICSjdbcDataSource: First Select Statement created
**** CICSjdbcDataSource: First Result Set created
**** CICSjdbcDataSource: Table name = SYSCOPY
**** CICSjdbcDataSource: Table name = SYSCOLAUTH
**** CICSjdbcDataSource: Table name = SYSCOLUMNS
.
. a list of table names defined to DB2
.
**** CICSjdbcDataSource: Table name = SYSSEQUENCESDEP
**** CICSjdbcDataSource: Result Set output completed
**** CICSjdbcDataSource: ended
*********************************************************************************
Figure 31. Stdout output from transaction DSDB to look up a DataSource with default name and subContext
You should be aware that if you do this, the code will no longer be portable in the
true sense, as it is now platform and vendor specific.
| If the DataSource has not been published or has already been retracted, the
| following message is displayed on the screen:
| DSRE - CICSDataSourceRetract: Datasource jdbc/CICSDB2DataSource not found
Figure 32. Stdout output from transaction DSRE to retract a DataSource with default name and subContext
A JDBC or SQLJ application can also issue JCICS commit or rollback directly, and
this has the same result as issuing a JDBC or SQLJ commit or rollback method
call. The whole unit of work is committed or rolled back together, both DB2 updates
and updates to CICS controlled resources.
When you are working with JDBC connections, there are some circumstances in
which you cannot avoid a syncpoint being taken, and the unit of work being
committed, when the connection to the DB2 database is closed. This applies in
either of the following circumstances:
v You have used the autocommit property of a JDBC connection. (See
“Autocommit.”)
v You have acquired the connection using an explicit URL. (See “Syncpoint issues
for explicit and default URLs” on page 130.)
Autocommit
JDBC applications can use the autocommit property of a JDBC connection. The
autocommit property causes a commit after each update to DB2. This commit is a
CICS commit, and results in the whole unit of work being committed.
Chapter 8. Using JDBC and SQLJ to access DB2 data from Java programs and enterprise beans written for CICS 129
Using the autocommit property also causes a commit to be taken when a
connection is closed, both for connections obtained using an explicit URL, and
connections obtained using a default URL.
The use of autocommit in a CICS environment is not recommended, and for this
reason the DB2 JDBC driver sets a default of autocommit(false) when running in a
CICS environment, which differs from non-CICS environments where the default is
autocommit(true).
You can overcome this restriction by acquiring the connection using a default URL
instead of an explicit URL, or by using a DataSource that provides a default URL
connection (see “Acquiring a connection to a database” on page 122). When a
default URL is used, the Java application does not have to close the connection on
a unit of work boundary, and no syncpoint is taken when the connection is closed
(provided that autocommit(true) has not been specified).
When using JDBC and SQLJ in enterprise beans, bear in mind the following
considerations:
v As for all Java programs in the CICS environment, it is better to use a default
URL, rather than an explicit URL, when obtaining a JDBC connection or an SQLJ
connection context (see “Committing a unit of work” on page 129 for more
information).
v For enterprise beans, it is particularly important to avoid using the autocommit
property of a JDBC connection. Autocommit(true) must not be used in an
enterprise bean running as part of an OTS transaction; this causes an ASPD
abend, because it causes a CICS syncpoint to be taken, which is not allowed in
an OTS transaction. For the same reason, do not use EXEC CICS SYNCPOINT,
JCICS commit, or JDBC or SQLJ commit commands in any program or
enterprise bean running as part of an OTS transaction.
v Remember that only one connection or connection context can be open at a
time. An enterprise bean using JDBC or SQLJ should close the JDBC connection
or SQLJ connection context before invoking methods on another enterprise bean
that also wishes to use JDBC or SQLJ within the same CICS transaction context.
Using a default URL ensures that no syncpoint is taken when the close occurs.
v DROLLBACK(YES) should not be specified on a DB2ENTRY definition or the
DB2CONN pool definition used by transactions running enterprise beans as part
of an OTS transaction. With this attribute, if a deadlock is detected, the CICS
DB2 attachment facility issues a CICS syncpoint rollback request, which is not
allowed in an OTS transaction, and an ASPD abend results. Enterprise beans
should use DROLLBACK(NO), and test for an SQLException with an SQLCODE
of —913 and issue an OTS rollback request.
Chapter 8. Using JDBC and SQLJ to access DB2 data from Java programs and enterprise beans written for CICS 131
132 CICS TS for z/OS: CICS DB2 Guide
Chapter 9. Preparing CICS DB2 programs for execution and
production
This chapter discusses program preparation in a CICS DB2 environment:
v “The CICS DB2 test environment”
v “CICS DB2 program preparation steps” on page 134
v “What to bind after a program change” on page 137
v “Bind options and considerations for programs” on page 138
v “CICS DB2 program testing and debugging” on page 139
v “Going into production: checklist for CICS DB2 applications” on page 139
v “Tuning a CICS application that accesses DB2” on page 142
For information on support for Java programs and enterprise beans in the CICS
DB2 environment, see Chapter 8, “Using JDBC and SQLJ to access DB2 data from
Java programs and enterprise beans written for CICS,” on page 117.
The first alternative, using a single CICS system for both production and test, is not
recommended. In releases of CICS before CICS Transaction Server for OS/390,
Version 1 Release 2, this was not recommended because the RCT could not be
dynamically changed. You can now change the CICS DB2 definitions using RDO
without stopping the attachment facility. Programs, transactions, maps, terminals
and DB2 definitions can all be dynamically added to a running system. However,
this environment is still not recommended, because applications in test could affect
the performance of the production system.
The second alternative, with just one DB2 system, could be used for both test and
production. Whether it is suitable depends on the development and production
environments involved. Running a test CICS system and a production CICS system
separately allows test failures without impacting production.
The third alternative, with, for example, one test and one production DB2 system, is
the most flexible. Two CICS subsystems can run with one or more DB2 systems.
Where the CICS systems are attached to different DB2 systems:
v User data and the DB2 catalog are not shared. This is an advantage if you want
to separate test data from production data.
v Wrong design or program errors in tested applications do not affect the
performance in the production system.
Figure 33. Steps to prepare a CICS application program that accesses DB2
For an overview of the stages in this process, see “Preparing a CICS application
program that accesses DB2” on page 11.
Table 9 shows the tasks that you need to perform to prepare a CICS DB2 program,
depending on the language of the program and on your version of DB2:
Table 9. Tasks to prepare a CICS program that accesses DB2
DB2 version and Step 1 Step 2 (CICS Step 3 Step 4 Step 5
program (SQL command (Program (Link-edit) (Bind)
language statement translation) compile)
processing)
DB2 V6 and DB2 CICS-supplied Language Link-edit with Bind
Assembler precompiler separate compiler EXEC interface process
translator and DSNCLI
Chapter 9. Preparing CICS DB2 programs for execution and production 135
Table 9. Tasks to prepare a CICS program that accesses DB2 (continued)
DB2 version and Step 1 Step 2 (CICS Step 3 Step 4 Step 5
program (SQL command (Program (Link-edit) (Bind)
language statement translation) compile)
processing)
DB2 V6 and PL/I DB2 Language compiler that Link-edit with Bind
precompiler supports integrated CICS EXEC interface process
translator and DSNCLI
DB2 V6 and DB2 Language compiler that Link-edit with Bind
COBOL precompiler supports integrated CICS EXEC interface process
translator and DSNCLI
DB2 V6 and DB2 CICS-supplied Language Link-edit with Bind
other languages precompiler separate compiler EXEC interface process
translator and DSNCLI
DB2 V7 and DB2 CICS-supplied Language Link-edit with Bind
Assembler precompiler separate compiler EXEC interface process
translator and DSNCLI
DB2 V7 and PL/I Language compiler that supports integrated Link-edit with Bind
CICS translator and SQL statement EXEC interface process
coprocessor and DSNCLI
DB2 V7 and Language compiler that supports integrated Link-edit with Bind
COBOL CICS translator and SQL statement EXEC interface process
coprocessor and DSNCLI
DB2 V7 and DB2 CICS-supplied Language Link-edit with Bind
other languages precompiler separate compiler EXEC interface process
translator and DSNCLI
You can perform this program preparation using the DB2 Interactive Interface
(DB2I) or by submitting your own JCL for batch execution.
v DB2 Interactive Interface (DB2I): DB2I provides panels to precompile, compile or
assemble, and link-edit an application program and to bind the plan. For details
about application program preparation, see the DB2 Universal Database for
OS/390 and z/OS Application Programming and SQL Guide.
v User JCL submitted to batch execution: Members DSNTEJ5C and DSNTEJ5P in
the DB2 library, SDSNSAMP, contain samples of the JCL required to prepare
COBOL and PL/I programs for CICS.
If you perform this process while CICS is running, you may need to issue a CEMT
NEWCOPY command to make the new version of the program known to CICS.
With DB2 Version 3 Release 1, DSNTIAR was split into two front-end modules
(DSNTIAC and DSNTIAR) and a run-time module (DSNTIA1). DSNTIAC is used for
CICS applications and DSNTIAR for other DB2 interfaces. This change removed
the need, previous to DB2 3.1, to relink-edit your application modules every time a
change is made to DSNTIAR, either by change of release or by applying
maintenance. If you have applications that have previously been link-edited with
DSNTIAR, you should consider link-editing them again using DSNTIAC instead,
which will provide performance improvments and isolate them from changes to
DSNTIAR.
The necessary program definitions for DSNTIAC and DSNTIA1 are provided in IBM
supplied group DFHDB2 on the CSD. You must add the SDSNLOAD library to the
CICS DFHRPL concatenation (after the CICS libraries) so that DSNTIA1 can be
loaded.
Module A
Module B Module C
Module D
Assuming that at least one SQL statement changed in program C, you must
perform the following steps to prepare the program and to make the transaction
executable again:
1. Precompile the program on DB2.
2. Translate the program using the CICS translator.
3. Compile the host language source statements.
4. Link-edit.
5. If the DBRM for program C was bound into a package, bind that package
using the new DBRM, and all the application plans that use program C will
automatically locate the new package.
6. If the DBRM for program C was bound directly into any application plans,
locate all the application plans that include the DBRM for program C. Bind all
the application plans again, using the DBRMs for all the programs directly
bound into them, to get new application plans. For the programs that were not
changed, use their old DBRMs. Note that you cannot use the REBIND
subcommand, because input to REBIND is the plan and not the DBRMs.
If you have not used packages before, note that using packages simplifies the
rebinding process. You can bind each separate DBRM as a package and include
them in a package list. The package list can be included in a PLAN. You can then
use the BIND PACKAGE command to bind the DBRMs for any changed programs,
Chapter 9. Preparing CICS DB2 programs for execution and production 137
instead of using the BIND PLAN command to bind the whole application plan. This
provides increased transaction availability and better performance. See section
“Using packages” on page 90 for more information on using packages.
When binding multiple programs into an application plan, be aware of the way in
which DB2 uses time stamps. For each program, the DB2 precompiler:
v Creates a DBRM with a time stamp of Tdx (for example Td1 for the first program,
Td2 for the second program, and so on).
v Creates a modified source program with a time stamp of Tsx in the SQL
parameter list (for example Ts1 and Ts2, if two programs are involved).
At bind time, the DBRM for each program is bound into the package or plan that
you have specified. In addition, DB2 updates its catalog table SYSIBM.SYSDBRM
with one line for each DBRM, together with its time stamp. At execution time, DB2
checks the time stamps for each SQL statement, and returns a -818 SQL code if
the time stamp for the DBRM and the time stamp it has placed in the source
program are different (in our example, if Td1 and Ts1 are different, or Td2 and Ts2
are different). To avoid -818 SQL codes, use one of the following strategies:
v Bind all programs into packages, and list these packages in the application plan.
When a program changes, simply precompile, compile, and link-edit the program,
and bind it into a package again.
v If you bind any programs directly into application plans, ensure that for every new
or changed program, you precompile, compile, and link-edit the program, then
bind all the application plans that involve that program, using the DBRMs from all
the programs directly bound into those plans. Use the BIND command, not the
REBIND command, to do this.
When you bind a plan, a number of options are available. Almost all bind options
are application dependent and should be taken into account during the application
design. You should develop procedures to handle different BIND options for different
plans. Also, the procedures should be able to handle changes in BIND options for
the same plan over time.
The following sections describe some specific recommendations for BIND options
with CICS:
RETAIN
RETAIN® means that BIND and EXECUTE authorities from the old plan are not
changed.
When the RETAIN option is not used, all authorities from earlier GRANTs are
REVOKED. The user executing the BIND command becomes the creator of the
plan, and all authorities must be reestablished by new GRANT commands.
This is why it is recommended that you use the RETAIN option when binding your
plans in the CICS environment.
Note that the isolation level is specified for the complete plan. This means that if
RR is necessary for a specific module in CICS, then all the DBRMs included in the
plan must also use RR.
Also, if for performance reasons you decide to group a number of infrequently used
transactions together to use the same DB2ENTRY and let them use a common
plan, then this new plan must also use RR, if just one of the transactions requires
RR.
Note that using dynamic SQL does not require VALIDATE(RUN). Nevertheless,
dynamic SQL implies that a statement is bound at execution.
These tasks are highly dependent on the standards you have used in the test
system. For example, the tasks to be performed are different if:
v There are separate DB2 systems for test and production
Chapter 9. Preparing CICS DB2 programs for execution and production 139
v Only one DB2 is used for both test and production.
The following discussion assumes that you use separate DB2 and CICS
subsystems for test and production.
If you want to bind your programs directly into application plans, or if you
want to bind the programs into packages on the production system, you
must perform the bind process on the DBRMs that you have placed on the
production system. If you are binding your programs directly into application
plans, you must then bind all the application plans on the production system
that involve those programs. See “The bind process” on page 12 for more
information on the bind process. Note that due to various factors, such as
the sizes of tables and indexes, comparing the EXPLAIN output between
test and production systems can be useless. Nevertheless, it is
recommended that you run EXPLAIN when you first bind a plan on the
production system, to check the DB2 optimizer decisions.
GRANT EXECUTE
You must grant users EXECUTE authority for the DB2 application plans on
the production system.
Tests Although no further tests should be necessary at this point, stress tests are
useful and recommended to minimize the occurrence of resource
contention, deadlocks, and timeouts and to check that the transaction
response time is as expected.
CICS definitions
To have new application programs ready to run, update the following RDO
definitions on the CICS production system.
v RDO transaction definitions for new transaction codes
v RDO program definitions for new application programs and maps
Chapter 9. Preparing CICS DB2 programs for execution and production 141
v SIT for specific DB2 requirements, if it is the first DB2-oriented
application going into production
v RDO DB2ENTRY and DB2TRAN definitions for the applications. RDO
DB2CONN definition if it is the first DB2-oriented application going into
production. When defining the new transactions and application plans in
the DB2ENTRY you can use unprotected threads to get detailed
accounting and performance information in the beginning. Later, you can
use protected threads as needed.
In addition, if RACF is installed, you need to define new users and DB2
objects.
When moving a CICS application that accesses DB2 to production, add these
checks to those already performed for CICS:
v If you are connecting CICS to DB2 Version 6 or later, check that all the
application programs that make DB2 requests are threadsafe. If they are, you will
be exploiting the open transaction environment (OTE), and improving the
performance of the application. See “Enabling CICS DB2 applications to exploit
the open transaction environment (OTE) through threadsafe programming” on
page 106 for an explanation of how application programs work in the open
transaction environment.
v Ensure that the number and type of SQL statements used meet the program
specifications (use the DB2 accounting facility).
v Check if the number of get and updated pages in the buffer pool is higher than
expected (use the DB2 accounting facility).
v Check that planned indexes are being used (use EXPLAIN), and that inefficient
SQL statements are not being used.
v Check if DDL is being used and, if so, the reasons for using it (use the DB2
accounting facility).
v Check if conversational transactions are being used.
Determine whether pseudoconversational transactions can be used instead. If
conversational design is needed, check the DB2 objects that are locked across
conversations. Check also that the number of new threads needed because of
this conversational design is acceptable.
v Check the locks used and their duration.
Make sure that tablespace locks are not being used because of incorrect or
suboptimal specification of, for example:
– LOCK TABLE statement
– LOCKSIZE=TS specification
– ISOLATION LEVEL(RR) specification
– Lock escalation.
This information is available in the catalog tables, except for lock escalation,
which is an installation parameter (DSNZPARM).
v Check the plans used and their sizes. Even though the application plans are
segmented, the more DBRMs used in the plan, the longer the time needed to
BIND and REBIND the plans in case of modification. Try to use packages
whenever possible. Packages were designed to solve the problems of:
When this tuning is complete, use the expected transaction load to decide on the
DB2ENTRY definitions required, and the number of threads required. Check also
the impact of these transactions on the DB2 and CICS subsystems.
Chapter 9. Preparing CICS DB2 programs for execution and production 143
144 CICS TS for z/OS: CICS DB2 Guide
Chapter 10. Accounting and monitoring in a CICS DB2
environment
This chapter deals with accounting and monitoring in a CICS DB2 environment. It
covers:
v “CICS-supplied accounting and monitoring information”
v “DB2-supplied accounting and monitoring information” on page 146
v “Monitoring a CICS DB2 environment: Overview” on page 147
v “Accounting in a CICS DB2 environment: Overview” on page 156
For advice on tuning the CICS DB2 attachment facility, see the CICS Performance
Guide.
For a full description of the CICS monitoring facilities, and details on activating,
collecting, and processing this information, see the CICS Performance Guide, and
the CICS Customization Guide.
For both statistics data and monitoring data, you can use an offline processing
facility. CICS Performance Analyzer and Tivoli® Decision Support for OS/390 are
two tools that collect and analyze data from CICS and other IBM systems and
products. They can build reports that help you with:
v Systems overview
v Service levels
v Availability
v Performance and tuning
v Capacity planning
The DB2 statistics trace can be activated for several classes. If the
statistics records are written to SMF, the SMF types are 100 and 102.
Accounting
Describes the work performed on behalf of a particular user (authorization
ID from the DB2CONN or DB2ENTRY). The main purposes of the
accounting records are to charge the DB2 cost to the authorization ID and
perform monitoring and tuning at the program level. DB2 produces an
accounting record at thread termination or when a transaction is reusing a
thread with a new authorization ID. That means that if a thread is defined
as protected (PROTECTNUM>0) and all transactions with the same
transaction code for this DB2ENTRY use the same authorization ID, only
one accounting record is produced, describing all activity done in the
thread. Additionally, accounting records are written if you set
ACCOUNTREC in your DB2ENTRY or DB2CONN definitions to UOW,
TASK, or TXID. Setting ACCOUNTREC to these options is considered a
signon, even if you use the same authorization ID.
You can activate the DB2 accounting trace for several classes. If the
accounting records are written to SMF, the SMF type is 101 and 102.
Audit Collects information about DB2 security controls and is used to ensure that
data access is allowed only for authorized purposes. If the audit records are
written to SMF, the SMF type is 102.
Performance
Records information for a number of different event classes. The
information is intended for:
v Program-related monitoring and tuning
v Resource-related monitoring and tuning
v User-related monitoring and tuning
v System-related monitoring and tuning
v Accounting-related profile creation
Include monitoring in the acceptance procedures for new applications, so that any
problems not detected during the test period can be quickly identified and corrected.
You can use some, or all, of the following tools to monitor the CICS DB2
attachment facility and CICS transactions that access DB2 resources. You can:
v Monitor the CICS DB2 attachment facility using:
– CICS DB2 attachment facility commands
– DB2 commands
– CICS DB2 statistics
See “Monitoring the CICS DB2 attachment facility” on page 148.
v Monitor CICS transactions using:
– CICS monitoring facility (CMF)
– CICS auxiliary trace
See “Monitoring CICS transactions that access DB2 resources” on page 151.
v Monitor DB2 using:
For example, the DSNC -DIS THREAD command can show CICS DB2 threads.
The command is routed to DB2 for processing. DB2 checks that the authorization
ID passed from CICS is authorized to issue the command entered.
Responses are routed back to the originating CICS user. The command recognition
character (CRC) of a hyphen, (-), must be used to distinguish DB2 commands from
CICS DB2 attachment facility commands. For DB2 commands issued from CICS,
the CRC is always -, regardless of the subsystem recognition character.
Both CICS and DB2 authorization are required to issue DB2 commands from a
CICS terminal:
v CICS authorization is required to use the DSNC transaction, and
v DB2 authorization is required to issue DB2 commands.
For more information see Chapter 4, “CICS-supplied transactions for CICS DB2,” on
page 33.
Monitoring the CICS DB2 attachment facility using CICS DB2 statistics
In addition to the limited statistics output by the DSNC DISP STAT command and
those output to the STATSQUEUE destination of the DB2CONN during attachment
The CICS DB2 global and resource statistics are described in detail in the CICS
Performance Guide.
CICS DB2 statistics are supported for all types of CICS statistics, namely:
v Requested statistics - CICS DB2 statistics are written as a result of an EXEC
CICS PERFORM STATISTICS RECORD command with the DB2 keyword.
v Requested reset statistics - a special case of requested statistics in which the
statistics counters are reset after collection.
v Interval statistics - statistics written when a requested interval expires.
v End of day statistics - a special case of interval statistics.
v Unsolicited statistics - CICS writes DB2 global and resource statistics to SMF
when the attachment facility is shut down. Also, DB2 resource statistics are
written to SMF when a DB2ENTRY is discarded.
The CICS sample statistics program, DFH0STAT, supports DB2 statistics. It uses
EXEC CICS COLLECT STATISTICS commands with the DB2CONN and
DB2ENTRY keywords to collect statistics. It also uses EXEC CICS INQUIRE
commands for the DB2CONN and DB2ENTRYs to collect data. An example of the
output from DFH0STAT is shown in Figure 35 on page 150.
DB2 Connection
DB2 Connection Status. . . . . . . . . . : CONNECTED DB2 Connect Date and Time . . . : 09/09/2001 10:37:19.21354
DB2 Connection Error . . . . . . . . . . : SQLCODE
DB2 Standby Mode . . . . . . . . . . . . : RECONNECT
Figure 35. Example output from DFH0STAT: the DB2 Connection report
DB2 Entries
Figure 36. Example output from DFH0STAT: the DB2 Entries report
CICS monitoring is used in the CICS DB2 environment with the DB2 accounting
facility, to monitor performance and to collect accounting information.
When CICS is connected to DB2 Version 6 or later, and is exploiting the open
transaction environment, the data in CICS performance records includes all
processor consumption for the period when the transaction is using DB2 resources.
You do not need to take data from DB2 into account when monitoring total
processor consumption. When CICS is connected to DB2 Version 5 or earlier, the
data in CICS monitoring records does not include the period when the transaction is
using DB2 resources, so you need to correlate the data from CICS and from DB2 to
obtain an accurate picture of processor consumption. For more information about
matching up CICS performance class records and DB2 accounting records, see
“Relating DB2 accounting records to CICS performance class records” on page
161. For more information about calculating processor consumption, see
“Accounting for processor usage in a CICS DB2 environment” on page 167.
You can use the CICS auxiliary trace facility to trace SQL calls issued by a CICS
application program.
For more information about trace output by the CICS DB2 attachment facility, see
Chapter 11, “Problem determination for CICS DB2,” on page 177.
The reports in this topic are shown as examples. Refer to the documentation of the
DB2PM release you are using for the format and meaning of the fields involved in
the reports.
These statistics are useful for tuning the DB2 subsystem, since they reflect the
activity for all subsystems connected to DB2.
It is difficult to interpret this data when more than one subsystem is connected to
DB2 (that is, both CICS and TSO). However, the counts obtained while running the
CICS DB2 attachment facility in a controlled environment (that is, with CICS as the
only subsystem connected, or with limited TSO activity) can be very useful.
The DB2 Universal Database for OS/390 and z/OS Administration Guide shows and
analyzes, from a DB2 viewpoint, the statistical data reported for the database and
system services address spaces. Included here is a reduced version of the statistics
report. You can use this report to monitor the average CICS transaction. Figure 37
on page 154 shows a small part of the report provided by DB2PM. Refer to the
DB2 Universal Database for OS/390 and z/OS Administration Guide for additional
information on these reports.
CPU TIMES TCB TIME SRB TIME TOTAL TIME /THREAD /COMMIT
------------------------------- --------------- --------------- --------------- --------------- ---------------
SYSTEM SERVICES ADDRESS SPACE 17:40.602755 1:09.182200 18:49.784954 51.353862 26.274069
DATABASE SERVICES ADDRESS SPACE 6.100449 11.626277 17.726726 0.805760 0.412249
IRLM 0.051894 2:43.867972 2:43.919867 7.450903 3.812090
DDF ADDRESS SPACE 1.195607 0.212343 1.407950 0.063998 0.032743
Further useful information in the statistics reports, not shown in Figure 37 on page
154, is:
v Locking can be used to monitor the number of timeouts and deadlocks. For more
information about deadlocks, see “Handling deadlocks in the CICS DB2
environment” on page 197.
v Buffer Pool provides information on:
– The number of data sets opened (Data sets Opened)
– The number of pages retrieved (GETPAGE Requests)
– Number of I/Os (Read Operations and Write I/O Operations)
This statistical data is not checkpointed and is not retained by DB2 across restarts.
Due to high resource consumption, the DB2 performance trace should be used only
in specific cases, where it becomes difficult to use any other tool to monitor
DB2-oriented transactions.
Even in this case, only the needed classes of performance trace should be started,
for only a limited time and for only the transactions that need to be carefully
monitored.
For example, data in the dispatcher statistics section provides the accumulated time
for each of the CICS TCBs. The field Accum/TCB is the total processor time used
by the corresponding TCB. For more information about the CICS Dispatcher
statistics, see the CICS Performance Guide.
You can obtain the total processor time used by the CICS address space from
RMF™ Monitor II reports. This time is usually greater than the sum of all CICS task
TCBs.
The difference between the processor time reported by RMF and the sum of CICS
TCBs in the CICS dispatcher statistics report is the processor time consumed by all
the other subtask TCBs. The subtasks are used for:
v DB2 threads (when CICS is connected to DB2 Version 5 or earlier)
v If DBCTL is used, processor time consumed in the DBCTL threads
v If MQSeries® is used, processor time consumed by the MQSeries threads
These are global performance reports and can help you determine how much of
your processor time is being used by CICS.
When planning the accounting strategy for your CICS DB2 environment, you need
to:
v Decide the types of DB2 accounting data to use in your accounting process
(processor usage, I/O, calls, and so on). “Accounting information provided by the
DB2 accounting facility” tells you about the accounting data that you can obtain
from the DB2 accounting facility.
v Decide how you are going to relate the data from the DB2 accounting record for
each transaction to the CICS performance class data for that transaction, to
create a complete picture of resource usage for the transaction. “Relating DB2
accounting records to CICS performance class records” on page 161 tells you
how you can match up the two types of data.
v Decide whether you are going to relate the CICS performance records and the
DB2 accounting records for each transaction back to the specific end user, or
whether you are going to define and calibrate a number of model transactions,
measure these transactions in a controlled environment, and count only the
number of model transactions executed by each end user. “Strategies you can
use to match DB2 accounting records and CICS performance class records and
charge resources back to the end user” on page 164 gives suggestions for when
each method is most appropriate.
If you have decided to use processor usage as the basis for your accounting,
“Accounting for processor usage in a CICS DB2 environment” on page 167 has
more information on the different classes of processor time that are reported in the
DB2 accounting records, and on how to calculate the total processor time used by a
transaction.
The identification section of each DB2 accounting record written to SMF and GTF
provides a number of keys on which the data can be sorted and summarized.
These include the authorization ID, the transaction ID, the plan name, and the
package name.
You have several possibilities for defining a cost formula based on the DB2
accounting records.
v Where repeatability combined with a reasonable expression for the complexity of
the transactions has high priority, then the processor usage, the GETPAGE
count, and the set write intents count are good candidates.
v If the purpose of the accounting process is to analyze the behavior of the CICS
transactions, then any information in the DB2 accounting records can be used.
Processor usage
The processor usage information given in the DB2 accounting record shows in most
cases the greater part of the total processor time used for the SQL calls. The DB2
statistics records report processor time used in the DB2 address spaces that could
not be related directly to the individual threads.
You should consider distributing the processor time reported in the DB2 statistics
records proportionally between all users of the DB2 subsystem (transactions, batch
programs, TSO users).
The amount of processor time reported in the DB2 accounting records is (for the
same work) relatively repeatable over time.
See “Accounting for processor usage in a CICS DB2 environment” on page 167 for
more detail on reporting processor usage in a CICS DB2 environment.
I/O
In a DB2 system, the I/O can be categorized in these types:
Synchronous read I/O
Sequential prefetch (asynchronous reads)
Asynchronous writes
EDM pool reads (DBDs and plan segments)
Log I/O (mainly writes).
Of these five I/O types, only the synchronous read I/O is recorded in the DB2
accounting record.
The number of sequential prefetch read requests is also reported, but the number
of read requests is not equal to the number of I/O.
None of the I/O types should be considered as repeatable over time. They all
depend on the buffer sizes and the workload activity.
GETPAGE
GETPAGE represents a number in the DB2 accounting record that is fairly constant
over time for the same transaction. It shows the number of times DB2 requested a
page from the buffer manager. Each time DB2 has to read or write data in a page,
the page must be available, and at least one GETPAGE is counted for the page.
This is true for both index and data pages. How often the GETPAGE counter is
incremented for a given page used several times depends on the access path
selected. However, for the same transaction accessing the same data, the number
of GETPAGEs remains fairly constant over time, but the GETPAGE algorithm can
change between different releases of DB2.
If the buffer pool contains the page requested, no I/O occurs. If the page is not
present in the buffer, the buffer manager requests the page from the media
manager, and I/O occurs.
The GETPAGE number is thus an indicator of the activity in DB2 necessary for
executing the SQL requests.
Write intents
The number of set write intents is present in the QBACSWS field of the DB2
accounting record, but the number is not related to the actual number of write I/Os
from the buffer pools. The number represents the number of times a page has been
marked for update. Even in a read-only transaction this number can be present,
because the intended writes to the temporary work files used in a DB2 sort are also
counted.
The typical case is that the number of set write intents is much higher than the
number of write I/Os. The ratio between these two numbers depends on the size of
the buffer pool and the workload. It is not a good measurement for write I/O activity,
but does indicate the complexity of the transactions.
A given SQL call can be simple or complex, depending on factors such as the
access path chosen and the number of tables and rows involved in the requests.
Transaction occurrence
A straightforward way of accounting is to track the number and type of transactions
executed. Your accounting is then based on these values.
Storage
The DB2 accounting record does not contain any information about real or virtual
storage related to the execution of the transactions. One of the purposes of the
DB2 subsystem is to optimize the storage use. This optimization is done at the DB2
level, not at the transaction level.
Because no information is given in the DB2 accounting record about the storage
consumption and because the storage use is optimized at the subsystem level, it is
difficult to account for storage in a DB2 environment.
AVERAGE APPL(CL.1) DB2 (CL.2) IFI (CL.5) CLASS 3 SUSPENSIONS AVERAGE TIME AV.EVENT HIGHLIGHTS
------------ ---------- ---------- ---------- -------------------- ------------ -------- --------------------------
ELAPSED TIME 25.435644 0.504442 N/P LOCK/LATCH(DB2+IRLM) 0.000000 0.00 #OCCURRENCES : 2
NONNESTED 25.435644 0.504442 N/A SYNCHRON. I/O 0.085908 6.50 #ALLIEDS : 2
STORED PROC 0.000000 0.000000 N/A DATABASE I/O 0.085908 6.50 #ALLIEDS DISTRIB: 0
UDF 0.000000 0.000000 N/A LOG WRITE I/O 0.000000 0.00 #DBATS : 0
TRIGGER 0.000000 0.000000 N/A OTHER READ I/O 0.042337 1.00 #DBATS DISTRIB. : 0
OTHER WRTE I/O 0.000000 0.00 #NO PROGRAM DATA: 2
CPU TIME 0.016663 0.015404 N/P SER.TASK SWTCH 0.352902 4.00 #NORMAL TERMINAT: 2
AGENT 0.016663 0.015404 N/A UPDATE COMMIT 0.000000 0.00 #ABNORMAL TERMIN: 0
NONNESTED 0.016663 0.015404 N/P OPEN/CLOSE 0.206822 1.50 #CP/X PARALLEL. : 0
STORED PRC 0.000000 0.000000 N/A SYSLGRNG REC 0.024259 1.00 #IO PARALLELISM : 0
UDF 0.000000 0.000000 N/A EXT/DEL/DEF 0.121821 1.50 #INCREMENT. BIND: 0
TRIGGER 0.000000 0.000000 N/A OTHER SERVICE 0.000000 0.00 #COMMITS : 3
PAR.TASKS 0.000000 0.000000 N/A ARC.LOG(QUIES) 0.000000 0.00 #ROLLBACKS : 0
ARC.LOG READ 0.000000 0.00 #SVPT REQUESTS : 0
SUSPEND TIME N/A 0.481147 N/A STOR.PRC SCHED 0.000000 0.00 #SVPT RELEASE : 0
AGENT N/A 0.481147 N/A UDF SCHEDULE 0.000000 0.00 #SVPT ROLLBACK : 0
PAR.TASKS N/A 0.000000 N/A DRAIN LOCK 0.000000 0.00 MAX SQL CASC LVL: 0
CLAIM RELEASE 0.000000 0.00 UPDATE/COMMIT : 0.00
NOT ACCOUNT. N/A 0.007891 N/A PAGE LATCH 0.000000 0.00 SYNCH I/O AVG. : 0.013217
DB2 ENT/EXIT N/A 35.00 N/A NOTIFY MSGS 0.000000 0.00
EN/EX-STPROC N/A 0.00 N/A GLOBAL CONT. 0.000000 0.00
EN/EX-UDF N/A 0.00 N/A FORCE-AT-COMMIT 0.000000 0.00
DCAPT.DESCR. N/A N/A N/P ASYNCH IXL REQUESTS 0.000000 0.00
LOG EXTRACT. N/A N/A N/P TOTAL CLASS 3 0.481147 11.50
Figure 38. Accounting long report for a CICS transaction accessing DB2 resources
#OCCURS #ROLLBK SELECTS INSERTS UPDATES DELETES CLASS1 EL.TIME CLASS2 EL.TIME GETPAGES SYN.READ LOCK SUS
PLANNAME #DISTRS #COMMIT FETCHES OPENS CLOSES PREPARE CLASS1 CPUTIME CLASS2 CPUTIME BUF.UPDT TOT.PREF #LOCKOUT
--------------------------- ------- ------- ------- ------- ------- ------- -------------- -------------- -------- -------- --------
DSNJDBC 1 0 0.00 0.00 0.00 0.00 1.706541 1.003194 249.00 13.00 0.00
0 2 4.00 2.00 2.00 2.00 0.027471 0.025984 0.00 5.00 0
TESTP05 2 0 0.00 0.50 0.00 0.00 33.283119 0.215656 7.00 0.00 0.00
0 2 1.50 0.50 0.00 0.00 0.001908 0.001389 1.00 0.00 0
Figure 39. Accounting short report for a CICS transaction accessing DB2 resources
If you are examining resource usage because you are carrying out performance
analysis, rather than accounting, then you always need to match up the DB2
accounting records and the CICS performance class records. However, if you are
examining resource usage for accounting purposes, you might not need to match
up the DB2 accounting records and the CICS performance class records. You do
not need to match up the records if:
v You are using DB2 Version 6 or later
and
v You have chosen to use processor time consumption as the basis for your
accounting.
In this situation, the processor time consumed in DB2 is reported by the CICS
performance class records as well as by the DB2 accounting records, so the CICS
performance class record for a transaction gives you all the information on
processor time that you need to charge the resources for that transaction back to
the end user. If you are in this situation, skip the rest of this section, and instead
read the sections “Accounting for processor usage in a CICS DB2 environment” on
page 167 (to understand how processor time consumption is reported), and
“Calculating CICS and DB2 processor times for DB2 Version 6 or later” on page
174 (to find out how to use the information on processor time that is available to
you).
If you are using DB2 Version 5 or earlier, or you have chosen to use data as well
as, or other than, processor time consumption as the basis for your accounting,
read the rest of this section to find out how to match up the DB2 accounting records
and the CICS performance class records. If you are using processor time
What are the issues when matching DB2 accounting records and CICS
performance records?
Because CICS and DB2 have different accounting needs, it is not always easy to
match up DB2 accounting records and CICS performance class records. There are
two main issues involved:
1. There is not necessarily a one-to-one relationship between the CICS
performance class records and the DB2 accounting records. A DB2 accounting
record can contain information about one CICS transaction, multiple CICS
transactions, or part of a CICS transaction.
2. The DB2 accounting records do not have a field that matches exactly with the
corresponding CICS performance records.
For the purpose of charging resources back to end users, it is possible to give each
end user a different authorization ID from a DB2 viewpoint, by specifying the
DB2ENTRY or DB2CONN parameter AUTHTYPE as OPID, USERID, GROUP, or
TERM. In this case, a DB2 accounting record is generated that contains data only
for the authorization ID. You can then collect together all the DB2 accounting
records by authorization ID, and charge the resources consumed directly to the end
user. This method means that you do not need to match the DB2 accounting
records with the CICS performance class records. However, from a usability and
performance viewpoint, using OPID, USERID, GROUP, and TERM is not an
attractive solution, for the reasons discussed in “Controlling users’ access to plans”
on page 84. For large networks, specifying these authorization IDs can complicate
maintenance and result in performance overhead. It is preferable to plan your use
of authorization IDs with performance in mind, and assign DB2 accounting records
to the end user by matching them to the CICS performance class records.
Even if you specify ACCOUNTREC(TASK), note that DB2 can only recognize a
single CICS task as long as the task continues to use the same thread. If a
transaction contains more than one UOW, assuming that it releases the thread at
the end of the UOW, it could use a different thread for each of its UOWs. This can
happen with terminal-oriented transactions issuing multiple syncpoints (commit or
rollback), and also non-terminal-oriented transactions if NONTERMREL(YES) is set
in the DB2CONN. In these cases, DB2 produces an accounting record for each
UOW, because it does not recognize them as a single task. So for this kind of
transaction, each DB2 accounting record can contain information about only a part
of the transaction, and you need to ensure that all the relevant DB2 accounting
records for the transaction are identified.
Strategies you can use to match DB2 accounting records and CICS
performance class records and charge resources back to the end user
There is not a single ideal way of matching DB2 accounting records and CICS
performance class records. In a few cases, it might be impossible to make the
matching correct, because transactions are being run concurrently. In most
situations, though, there are strategies you can use to match up the two types of
records with reasonable accuracy.
If the resources used in the individual transaction are the basis for accounting, then
when you have matched up the CICS performance records and the DB2 accounting
records, you can relate them back to the specific end user. Alternatively, you can
define and calibrate a number of model transactions, measure these transactions in
a controlled environment, and count only the number of model transactions
executed by each end user.
The two main factors that determine what strategies you should use are:
v Whether each CICS transaction ID represents only one possible transaction path
(and so it always represents the same amount of resources consumed), or
whether many different transaction paths share the same CICS transaction ID (so
it can represent different amounts of resources consumed.
v Whether each DB2 accounting record relates to only one transaction or a part of
one transaction (because you have taken one of the measures described in
“Controlling the relationship between DB2 accounting records and CICS
performance class records” on page 162), or whether it contains information
about more than one transaction.
Figure 40 on page 165 shows how these factors combine to create four typical
scenarios that you might encounter when matching DB2 accounting records and
CICS performance class records. The following sections suggest strategies for
matching the records in each case, and for charging the resources used back to the
end user.
Each transaction
path has its own A B
transaction ID
Many different
transaction paths C D
share the same
transaction ID
The end user for the transaction can be identified from the CICS performance
record. This record contains the CICS activities related to this transaction. You can
identify the DB2 accounting records that apply to this transaction by using any of
the data items listed in “Using data in the DB2 accounting record to identify the
corresponding CICS performance class records” on page 163.
As all these transactions are identical, you can expect that they consume
comparable amounts of resources. For accounting purposes, you could create
model transactions for each transaction type. Because you can identify which DB2
accounting records apply to which CICS transactions, you can match up the DB2
accounting records and the CICS performance record for one transaction, and then
simply assign the amount of DB2 resources used in those accounting records, to
each subsequent transaction of that type. You should validate the correctness of
your models on a regular basis, in case the resource usage changes.
If two or more different types of CICS transaction are present in a particular DB2
accounting record (because they use the same DB2ENTRY and hence the same
thread), you cannot use the method of distributing the resources equally, because
the different types of transaction might use different resources. In this case, you can
create model transactions by periodically measuring the amount of DB2 resources
used by each type of CICS transaction. Take these measurements by temporarily
disallowing thread reuse, and looking at the resulting DB2 accounting records,
which will contain information relating to only one transaction. Use these model
transactions to charge back the resources to the end user. You should periodically
validate the correctness of the model transactions.
You won’t be able to match up a set of records for one instance of a transaction
and then re-use those figures, as you could in Scenario A. You will need to match
all the individual CICS performance records with their corresponding DB2
accounting records. Unless you do this, you won’t know what type of transaction is
represented by each DB2 record.
You can match each of the DB2 accounting records to the relevant CICS
performance record by using the data items listed in “Using data in the DB2
accounting record to identify the corresponding CICS performance class records” on
page 163. If you have specified either ACCOUNTREC(TASK) or
ACCOUNTREC(UOW) in the DB2ENTRY or DB2CONN, so that CICS passes its
LU6.2 token to DB2, then you can match the records together easily. If not, you will
need to match the records based on their time stamps. In this case, the matching
may not be accurate if transactions are being run simultaneously.
You can then use your matched sets of records to charge back the resources used
for each transaction, to the end user identified by the CICS performance record.
This situation is best avoided, because you are unlikely to be able to match records
accurately. If you do find yourself in this situation, the best solution is to create
model transactions, as described for Scenario B. Next, find a way to mark the CICS
The DB2 accounting trace can be started with CLASS 1, CLASS 2, or CLASS 3.
However, CLASS 1 must always be active to externalize the information collected
by activating CLASS 2, CLASS 3, or both classes.
The processor times reported in the DB2 accounting records are the TCB time for
the thread TCB running code in CICS or in the DB2 address space, using
cross-memory services; and the SRB time for work scheduled in CICS.
CLASS 1 (the default) results in accounting data being accumulated by several DB2
components during normal execution. This data is then collected to write the DB2
accounting record. The data collection does not involve any overhead of individual
event tracing.
CLASS 2 and CLASS 3 activate many additional trace points. Every occurrence of
these events is traced internally, but is not written to an external destination. Rather,
the accounting facility uses these traces to compute the additional total statistics
that appear in the accounting record when CLASS 2 or CLASS 3 is activated.
Accounting CLASS 1 must be active to externalize the information.
CLASS 2 collects the delta elapsed and processor times spent ‘IN DB2’ and records
this in the accounting record.
CLASS 3 collects the I/O elapsed time and lock and latch suspension time spent ‘IN
DB2’ and records this in the accounting record.
CLASS 7 and CLASS 8 in DB2 collect package level accounting in DB2 and
package level accounting wait in DB2. For information on package level accounting,
refer to the DB2 Universal Database for OS/390 and z/OS Administration Guide.
The statistics trace reports processor time in the statistics records. The processor
times reported are:
v Time under a DB2 address space TCB running asynchronous of the CICS
address space. Examples of this are the DB2 log and writes from the buffer
pools.
v Time under SRBs scheduled under the DB2 address spaces. An example is the
asynchronous read engine for sequential prefetch.
The DB2 address spaces reported in the statistics record are:
v Database manager address space
v System services address space
v IRLM.
In a CICS DB2 environment, the processor time from the DB2 accounting records is
typically much greater than the processor time reported in the DB2 statistical
Figure 41 on page 169 and Figure 42 on page 171 show each period of processor
time that is reported by CICS and DB2, and where it takes place. The location of
processing differs depending on the version of DB2 to which CICS is connected,
and whether or not the application accessing DB2 is threadsafe. This is because
when CICS is connected to DB2 Version 6 or later, and is exploiting the open
transaction environment, the CICS DB2 attachment facility uses CICS-managed
open TCBs rather than CICS DB2 subtask TCBs.
Figure 41 on page 169 shows the situation when CICS is connected to DB2 Version
5 or earlier, or when the application accessing DB2 is not threadsafe.
CICS Thread
main TCB TCB
(QR TCB) (Subtask TCB)
3 2
6 5
9 8
10
Figure 41. CICS with DB2 Version 5 or non-threadsafe application: Processor times recorded
Figure 42 on page 171 shows the situation when CICS is connected to DB2 Version
6 or later, and the application accessing DB2 is threadsafe.
CICS Thread
main TCB TCB
(QR TCB) (Open TCB)
3 2
6 5
9 8
10
11
Figure 42. CICS with DB2 Version 6 and threadsafe application: Processor times recorded
Here, CICS is using an open TCB to run the thread into DB2. As the application is
threadsafe, the application code can also run on the open TCB. The periods of
processor time shown in the figure are as follows:
v Time 1: The application starts on the CICS main TCB. At the end of Time 1, the
application issues an EXEC SQL request.
v Time 2: DB2 is fulfilling the application’s request. This processor time is spent in
the DB2 address space. At the end of Time 2, DB2 passes its response to the
CICS DB2 attachment facility.
v Time 3: The CICS DB2 attachment facility is carrying out processing on the
thread TCB. This covers both the processing needed for the application to
access DB2, and the processing needed to pass DB2’s response back to the
application. It also includes the time spent waiting for DB2 to fulfil the request
and respond. Time 2 is therefore nested within Time 3. At the end of Time 3,
DB2’s response is passed to the application, and the access to DB2 is complete.
For CLASS 1, a task processor timer is created when the TCB is attached. When a
thread to DB2 starts, the timer value is saved. When the thread is terminated (or
the authorization ID is changed), then the timer is checked again, and both the
timer start and end values are recorded in the SMF type 101 record (the DB2
accounting record). The fields in the SMF type 101 record used for accounting
CLASS 1 processor time are:
v QWACBJST for begin thread TCB time
v QWACEJST for end thread TCB time
v QWACBSRB for begin ASCB SRB time
v QWACESRB for end ASCB SRB time.
You can find a description of the contents of the DB2 accounting record in the DB2
Universal Database for OS/390 and z/OS Administration Guide. There is also the
description of accounting record fields in member DSNWMSGS, which is shipped in
SDSNSAMP.
When CICS is connected to DB2 Version 6 or later, when the CLASS 1 recording
becomes active for a thread, it is recording time spent on the L8 open TCB.
Because the L8 TCB is used for both CICS activity and DB2 activity, this includes
processor time spent in the CICS-DB2 attachment facility, including trace calls. It
also includes processor time spent running application code (if the application is
threadsafe) and threadsafe CICS commands on the open TCB. If a thread is
reused, the thread housekeeping processor time is also included in the CLASS 1
When CICS is connected to DB2 Version 6 or later, you do not need to use the
DB2 CLASS 1 processor time for accounting purposes. The processor time
recorded in the CICS SMF type 110 record is all that is required to give a complete
account of the processor time consumed by an application. This record includes the
time spent in the application code, the thread creation and termination costs, and
the time covered by the DB2 CLASS 1 processor time. For more information, see
“Calculating CICS and DB2 processor times for DB2 Version 6 or later” on page
174.
For accounting CLASS 2, the timer is checked on every entry and exit from DB2 to
record the ‘IN DB2’ time in the SMF type 101 record. In this case, it is the
difference that is stored in the record. The fields in the SMF type 101 record used
for accounting CLASS 2 processor time are QWACAJST for CICS attach TCB time
and QWACASRB for CICS ASCB SRB time.
For a description of the contents of the DB2 statistics records, see the DB2
Universal Database for OS/390 and z/OS Administration Guide.
The elapsed time (start and end times between the above defined points) is also
reported in the SMF type 101 record (QWACASC).
When CICS is connected to DB2 Version 5 or earlier, the CLASS 2 processor time
is not significantly different from the CLASS 1 processor time, in contrast to IMS™
and TSO. However, when CICS is connected to DB2 Version 6 or later, there can
be a significant difference between CLASS 1 and CLASS 2 processor time, as
there is for IMS and TSO. This is because when CICS is connected to DB2 Version
6 or later, the CLASS 1 processor time includes processor time spent in the
CICS-DB2 attachment facility, including trace calls, and also includes processor
time spent running threadsafe application code and threadsafe CICS commands on
the open TCB. All this processor time occurs in the CICS address space, and so is
not reported for accounting CLASS 2. The CLASS 2 processor time itself is not
affected by the open transaction environment.
To reduce the overhead, it was usually recommended that the CLASS 2 trace be
limited to trace data for plans and locations of specific authorization IDs. The
processor overhead is 0-5%, with 2% being typical.
Calculating CICS and DB2 processor times for DB2 Version 5 or earlier
When your CICS system is connected to DB2 Version 5 or earlier, the CICS
performance class record does not include processor time consumed in DB2. To
estimate the total processor time for a single transaction, you could add information
from the corresponding CICS performance record and DB2 accounting record (SMF
type 101 record). For information about matching CICS performance records with
DB2 accounting records, see “Relating DB2 accounting records to CICS
performance class records” on page 161.
You should take the CPU field from the CICS performance class record. Using the
DB2 accounting record, you can calculate the DB2 thread processor time (T1) as:
T1 = QWACEJST - QWACBJST
This calculates the CLASS 1 TCB processor time. The sum of the processor time
from the CICS CPU field and the calculated value of T1 is an expression for the
processor time spent in CICS and DB2.
Notes:
v The CLASS 2 processor time is part of the CLASS 1 processor time and should
not be added to the sum.
v If the CLASS 2 processor time is subtracted from the CLASS 1 processor time,
this gives an approximation of CPU utilization of the CICS DB2 attachment
facility. This is a useful tuning aid.
v The processor time used in the DB2 address spaces and recorded in the DB2
statistics records is not related to any specific thread. It can be distributed
proportionally to the CPU time recorded in the DB2 accounting records.
v Processor time used in the CICS address space under the subtask TCBs cannot
easily be distributed to the CICS performance records, because it includes the
processor times for the DB2 subtasks, which are already contained in the
calculated T1 value. It means that processor time used in subtasks other than
the thread subtasks is not included in the addition.
v Most of the processor time used in the thread TCB to create the thread is not
included in any DB2 accounting records associated with this thread, because this
processor time is spent before the thread is created.
v The capture ratio for CICS and DB2 should be taken into account. Capture ratio
is the ratio of reported CPU time to total used CPU time (for more information,
see the z/OS Resource Measurement Facility (RMF) Performance Management
Guide (SC28–1951)).
Calculating CICS and DB2 processor times for DB2 Version 6 or later
When CICS is connected to DB2 Version 6 or later, and is exploiting the open
transaction environment, the CICS DB2 attachment facility uses CICS-managed
When CICS is connected to DB2 Version 6 or later, do not add together the
processor time from the CICS records (SMF type 110 records) and the DB2
accounting records (SMF type 101 records) when calculating the total processor
time for a single transaction, because the DB2 processor time would then be
included twice. The total processor time for a single transaction is recorded in the
USRCPUT field in the CICS records (performance class data field 008 from group
DFHTASK). This field includes all processor time used by the transaction when it
was executing on any TCB managed by the CICS dispatcher. CICS-managed TCBs
include the QR, RO, CO, H8, J8, and L8 mode TCBs.
In the open transaction environment, the CICS L8 task processor time can also
include the cost of creating a DB2 thread. When CICS is connected to DB2 Version
5 or earlier, this time is unaccounted for in the CICS and DB2 SMF records. When
CICS is connected to DB2 Version 6 or later, if a transaction causes a DB2 thread
to be created, you can expect the total task processor time accounted for to be
higher than that accounted for by a CICS system running with earlier DB2 releases.
Correspondingly, if at the end of a transaction, the thread is terminated (because it
is unprotected and no other task is waiting to use it), then the cost of thread
termination is included in the CICS L8 task processor time. Again, this cost is not
accounted for by a CICS system connected to DB2 Version 5 or earlier.
The capture ratio for CICS and DB2 should be taken into account. Capture ratio is
the ratio of reported CPU time to total used CPU time. For more information, see
the OS/390: Resource Measurement Facility Performance Management Guide
(SC28–1951).
When CICS is connected to DB2 Version 5 or earlier, the thread TCBs are
specially created “daughter” subtasks of a CICS DB2 attachment facility subtask
(the MSUB TCB) that is established when CICS connects to DB2. The MSUB TCB
is a subtask of the main CICS TCB (the QR TCB), and hence the thread TCBs are
“grand daughters” of the main CICS TCB. The CICS DB2 attachment facility runs
on the QR TCB. It uses module DFHD2MSB, running on the MSUB TCB, to control
the subtask thread TCBs.
The number of subtask thread TCBs allowed is controlled using the TCBLIMIT
attribute of the DB2 connection definition (DB2CONN). A subtask thread TCB is not
terminated when the thread is terminated. A subtask thread TCB can be terminated
if:
v A CICS transaction is force purged from CICS and the thread is still active in
DB2. In this case, the subtask is terminated as a means of flushing the request
out of DB2. The current UOW in DB2 is backed out.
v The TCBLIMIT value of the DB2CONN is lowered. When a thread is terminated,
the CICS DB2 attachment facility recognises that the current number of TCBs
exceeds the TCBLIMIT, and terminates the TCB as well.
v CICS is indoubt as to the outcome of a UOW because it has lost contact with its
coordinator. Terminating the subtask causes DB2 to release the thread, but
maintain the UOW as indoubt and maintain its locks. The UOW is completed by
a later resynchronization when CICS reestablishes contact with its coordinator.
v The CICS DB2 attachment facility is shut down.
When CICS is connected to DB2 Version 6 or later, the thread TCBs are open
L8 mode TCBs. The open TCBs are “daughters” of the main CICS TCB (the QR
TCB). The CICS DB2 task-related user exit itself runs on the open TCB, as well as
using it to run the thread. The task-related user exit uses the CICS DB2 attach
© Copyright IBM Corp. 1997, 2003 177
module DFHD2D2 to invoke DB2 when it needs to acquire a thread. Another
module, DFHD2CO, running on a different TCB, deals with aspects of the overall
CICS DB2 connection, including identifying to DB2 and disconnecting CICS from
DB2.
The maximum number of open TCBs that can be running threads into DB2 at any
one time is controlled using the TCBLIMIT parameter of the DB2CONN. An open
TCB running a thread is not terminated when the thread is terminated. An open
TCB can be terminated if:
v A CICS transaction is force purged from CICS and the thread is still active in
DB2. In this case the TCB is terminated as a means of flushing the request out
of DB2. The current UOW in DB2 is backed out.
v CICS is indoubt as to the outcome of a UOW because it has lost contact with its
coordinator. Terminating the TCB causes DB2 to release the thread, but maintain
the UOW as indoubt and maintain its locks. The UOW is completed by a later
resynchronization when CICS reestablishes contact with its coordinator.
v The CICS dispatcher, where open TCBs are returned if they are not being used
by the CICS DB2 attachment facility, cleans up the unused open TCBs after a
period of time.
When CICS is connected to DB2 Version 5 or earlier, DFHD2EX1 runs on the CICS
main TCB (the QR TCB). It uses the WAIT_MVS function of CICS dispatcher
domain to put the running CICS task into a CICS dispatcher wait while the DB2
request is run on a subtask TCB. This wait is represented by a resource type of
“DB2“.
When CICS is connected to DB2 Version 6 or later, the CICS DB2 task-related user
exit and the request into DB2 run on an L8 open TCB. The CICS task is not put into
a CICS dispatcher wait when active in DB2, so there is no related information in the
dispatcher section of the CICS system dump. The CEMT INQUIRE TASK panel
shows that the CICS DB2 task is running on an open TCB. To find out if the task is
active in DB2, use the DSNC DISPLAY TRAN command (see “DSNC DISPLAY” on
page 38). This command displays all the threads currently in use, and the task with
which each thread is associated. See “DISPLAY PLAN or TRAN” on page 39 for an
example of the output from the command. If the thread associated with the task has
a status of ‘*’ in the ‘S’ field, this shows that the thread is currently active in DB2.
The task is therefore either running or waiting in DB2.
As an example, take a situation that can occur when CICS is connected to DB2
Version 5 or earlier. When a CICS system dump is taken, a CICS task is waiting for
the CICS DB2 task to complete its work in DB2. This wait is represented by a
resource type of “DB2” and a resource name of “LOT_ECB”. The examples shown
in Figure 43 on page 179, Figure 44 on page 179, and Figure 45 on page 179 show
how this would appear in the dispatcher section of CICS system dump, and using
the CEMT inquire task panels.
INQUIRE TASK
STATUS: RESULTS - OVERTYPE TO MODIFY
Tas(0000151) Tra(DSNC) Sus Tas Pri( 255 )
? Tas(0000161) Tra(XC05) Fac(1303) Sus Ter Pri( 001 )
Tas(0000162) Tra(CEMT) Fac(1302) Run Ter Pri( 255 )
Figure 44. CICS CEMT INQUIRE command for a CICS DB2 wait transaction
INQUIRE TASK
SYNTAX OF SET COMMAND
Tas(0000161) Tra(XC05) Fac(1303) Sus Ter Pri( 001 )
Hty(DB2 ) Hva(LOT_ECB ) Hti(000018) Sta(TO)
Use(SYSADM ) Rec(X’A8B5FE112D54CA85’)
CEMT Set TAsk() | < All >
< PRiority() >
< PUrge | FOrcepurge >
Figure 45. CICS CEMT INQUIRE command after question mark (?)
The full list of waits issued from the CICS DB2 task-related user exit DFHD2EX1,
their meanings, and whether the task can be purged or forcepurged out of this wait
is given in Table 12. A fuller explanation of each wait follows the table.
Table 12. WAITs issued from the CICS DB2 TRUE
Resource type Resource Meaning Purge and
or Hty value name or Hva forcepurge
value
DB2 LOT_ECB CICS task is waiting for DB2, that is, Purge: No.
waiting for the CICS DB2 task to Forcepurge:
complete the request. Used when CICS Yes (but see
is connected to DB2 Version 5 or below).
earlier.
CDB2RDYQ name of CICS task is waiting for a thread to No
DB2ENTRY or become available. The resource name
*POOL details for which DB2ENTRY or the pool
has a shortage of threads.
A resource type of DB2 is only used when CICS is connected to DB2 Version 5 or
earlier. It indicates that the task is waiting for DB2 or, more specifically, the task is
waiting for its associated CICS DB2 task to complete the DB2 request and post it
back. The task can be forcepurged when in this state, in which case the CICS task
abends and backout occurs. In addition, the CICS DB2 task is detached, causing
DB2 to back out as well. Forcepurging a task in this state can cause termination of
the DB2 subsystem, if the CICS DB2 task is terminated during a “must complete”
activity in DB2. To avoid this risk, use the DB2 CANCEL THREAD command before
issuing the forcepurge—see “Purging CICS DB2 transactions” on page 30 for the
procedure.
Table 13 gives details of WAITS issued using WAIT_OLDC dispatcher calls where
the ECB is hand posted:
Table 13. WAITS issued using WAIT_OLDC dispatcher calls
Resource type Resource Meaning Purge and
or Hty value name or Hva forcepurge
value
DB2_INIT CICS DB2 initialization program Yes
DFHD2IN1 issues the wait to wait for
the CICS initialization task running
program DFHD2IN2 to complete.
DB2CDISC name of A SET DB2CONN NOTCONNECTED Yes
DB2CONN command has been issued with the
WAIT or FORCE option. DFHD2TM
waits for the count of tasks using DB2
to reach zero.
DB2EDISA name of A SET DB2ENTRY DISABLED Yes
DB2ENTRY command has been issued with the
WAIT or FORCE option. DFHD2TM
waits for the count of tasks using the
DB2ENTRY to reach zero.
Table 14 shows details of EXEC CICS WAIT EXTERNAL requests issued by the
CICS DB2 attachment facility.
Table 14. EXEC CICS WAIT EXTERNAL requests issued by the attachment facility
Resource type Resource Meaning Purge and
or Hty value name or Hva forcepurge
value
USERWAIT CDB2TIME The CICS DB2 service task program Yes
DFHD2EX2 is in its timer wait cycle
either waiting for the protected thread
purge cycle to pop or to be posted for
another event.
USERWAIT DB2START The CICS DB2 service program Yes
DFHD2EX2 is waiting for DB2 to post it
when it becomes active.
Where possible, message numbers have been maintained from previous releases
of the attachment. For example, message DFHDB2023 documents the same
condition as old messages DSN2023 or DSNC023. However, the contents of the
message have changed. For example, all messages contain the date, time, and
applid.
The destination for transient data messages output by the CICS DB2 attachment
facility is controlled using the MSGQUEUE1, MSGQUEUE2 and MSGQUEUE3
parameters of the DB2CONN definition. The same message can be routed to three
transient data queues. By default, messages are sent to a single queue called
CDB2 which is defined as indirect to queue CSSL.
All messages are documented in the CICS Messages and Codes manual and are
available using the CMAC CICS-supplied transaction.
Figure 46 on page 184 and Figure 47 on page 188 show examples of the trace
output from the RMI and the CICS DB2 task-related user exit, DFHD2EX1, when
level 1 and level 2 RI tracing is active. In both examples, the trace shows one SQL
statement being executed. The example trace in Figure 46 on page 184 was output
when CICS was connected to DB2 Version 5 or earlier. In this environment, the
CICS DB2 task-related user exit runs on the CICS QR TCB (the main TCB), and
creates subtask TCBs to run threads into DB2. The example trace in Figure 47 on
page 188 was output when CICS was connected to DB2 Version 6, and was using
the open transaction environment. In this environment, the CICS DB2 task-related
user exit runs on an open L8 mode TCB, and uses that same TCB to run threads
into DB2.
CICS DB2 trace output is written to the CICS internal trace table and auxiliary trace
and GTF trace destinations if active.
Figure 46. Sample trace output from the RMI and the CICS DB2 TRUE — CICS connected to DB2 Version 5 or earlier
(Part 1 of 4)
Figure 46. Sample trace output from the RMI and the CICS DB2 TRUE — CICS connected to DB2 Version 5 or earlier
(Part 2 of 4)
AP 3183 D2EX1 EVENT - AWOKEN BY D2EX3 THREAD TCB FOR DB2ENTRY XP05 USING PLAN TESTP05
Figure 46. Sample trace output from the RMI and the CICS DB2 TRUE — CICS connected to DB2 Version 5 or earlier
(Part 3 of 4)
Figure 46. Sample trace output from the RMI and the CICS DB2 TRUE — CICS connected to DB2 Version 5 or earlier
(Part 4 of 4)
Figure 47. Sample trace output from the RMI and the CICS DB2 TRUE — CICS connected to DB2 Version 6 or later
(Part 1 of 4)
Figure 47. Sample trace output from the RMI and the CICS DB2 TRUE — CICS connected to DB2 Version 6 or later
(Part 2 of 4)
5-0000 00DE6EC4 C6C8C4F2 D3D6E340 40404040 E7D7F0F5 184AC080 18DF2D78 18DCD030 *..>DFHD2LOT XP05.${.......}.*
0020 00000000 0006EFF0 00000000 18DF2E0C 18F0CDE0 00000000 00000000 FF6CD800 *.......0.........0.\.........%Q.*
0040 00000000 00000000 00000000 00000000 00000000 E3C5E2E3 D7F0F540 010C010C *....................TESTP05 ....*
0060 A000C000 00000000 00000000 00000000 C9E8D2F2 E9F2C7F1 B60AF030 A54A8EC0 *..{.............IYK2Z2G1..0.v$.{*
0080 D1E3C9D3 D3C9F140 40404040 40404040 00000000 00000000 C7C2C9C2 D4C9E8C1 *JTILLI1 ........GBIBMIYA*
00A0 C9E8C1D8 E3C3F0F3 0AF030A5 4A8EC6D9 C2400003 000118F0 CD680000 00000000 *IYAQTC03.0.v$.FRB .....0........*
00C0 00000000 00000000 39050001 04000000 00000000 00000000 00000000 0000 *.............................. *
Figure 47. Sample trace output from the RMI and the CICS DB2 TRUE — CICS connected to DB2 Version 6 or later
(Part 3 of 4)
Figure 47. Sample trace output from the RMI and the CICS DB2 TRUE — CICS connected to DB2 Version 6 or later
(Part 4 of 4)
When CICS is connected to DB2 Version 5 or earlier, the thread TCB is a subtask
TCB owned by the CICS DB2 attachment facility and not known to the CICS
dispatcher or kernel. This means that CICS tracing cannot be performed in the
subtask program, DFHD2EX3. DFHD2EX3 does issue one GTF trace to record the
point in time it posts back the CICS task on completion of the DB2 request. In
addition to this, DFHD2EX3 maintains its own trace table at the end of the CSUB
control block to record the requests it makes to DB2, and the responses to those
requests.
When CICS is connected to DB2 Version 6 or later, and is using the open
transaction environment, the thread TCB is an open TCB known to the CICS
dispatcher and kernel. The CICS DB2 thread processor DFHD2D2, which replaces
DFHD2EX3 in the open transaction environment, can therefore use CICS tracing.
As well as using CICS tracing, DFHD2D2, like DFHD2EX3, maintains a trace table
at the end of the CSUB control block to record the requests it makes to DB2, and
the responses to those requests.
The CSUB trace table is 160 bytes in length allowing ten entries of 16 bytes to be
written. The trace table wraps when all ten entries have been used. The format of
each trace entry is shown in Table 16.
Table 16. Layout of CSUB trace table entry
Bytes Content Information
Bytes 0-3 Trace request number Fullword number of the trace entry written. The number is used to
find the latest entry written.
Bytes 4-7 Trace request Four-character representation of the DB2 request issued. Possible
values are:
In the trace example Figure 47 on page 188, output when CICS is connected to
DB2 Version 5 or earlier, DATA6 from trace point 3181 is the CSUB, and this
section is shown in Figure 48. Looking at the character representation on the right
hand side, the trace table is delimited by >>Trace Start >> and <<Trace End <<. In
this example you can see that an identify, signon, create thread, followed by two
API requests have been issued to DB2, and all requests were successful and the
FRB return code and reason codes for each request are zeros.
6-0000 03006EC4 C6C8C4F2 C3E2C240 40404040 B60AF031 FAD0CB43 18D305E0 18DF2D78 *..>DFHD2CSB ..0..}...L.\....*
0020 19044210 007C6B90 00000000 00000000 B60AF030 A54A8EC0 00000000 00000000 *.....@,...........0.v$.{........*
0040 18DF2DE4 00000000 00000000 00000000 00000000 00000000 00000000 00000000 *...U............................*
0060 00000000 00000000 00000000 E3C5E2E3 D7F0F540 D1E3C9D3 D3C9F140 40404040 *............TESTP05 JTILLI1 *
0080 40404040 C5D5E3D9 E7D7F0F5 F0F0F0F1 00000000 B60AF032 01359743 C7C2C9C2 * ENTRXP050001......0...p.GBIB*
00A0 D4C9E8C1 C9E8C1D8 E3C3F0F3 0AF030A5 4A8E0000 44800000 00000001 00000000 *MIYAIYAQTC03.0.v$...............*
00C0 00000000 00000000 00000000 40404040 40404040 40404040 40404040 FF6CD800 *............ .%Q.*
00E0 C6D9C240 00030001 18F0CDE0 00000000 00000000 00000000 00003905 00010400 *FRB .....0.\....................*
0100 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 *................................*
0120 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 *................................*
0140 00000000 00000000 00000000 00000000 00000000 00000000 98DCD110 00000000 *........................q.J.....*
0160 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 *................................*
0180 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 *................................*
01A0 00000000 00000000 40404040 40404040 40404040 40404040 40404040 40404040 *........ *
01C0 40404040 40404040 40404040 40404040 40404040 40404040 40404040 00000000 * ....*
01E0 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 *................................*
0200 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 *................................*
0220 00000000 00000000 00000000 00000000 00000000 00000000 00000005 18DCD2C0 *..............................K{*
0240 6E6EE399 81838540 E2A38199 A3406E6E 0100035C C9C4C5D5 00000000 00000000 *>>Trace Start >>...*IDEN........*
0260 0200035C E2C9C7D5 00000000 00000000 0300035C C3E3C8C4 00000000 00000000 *...*SIGN...........*CTHD........*
0280 0400035C C1D7C940 00000000 00000000 0500035C C1D7C940 00000000 00000000 *...*API ...........*API ........*
02A0 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 *................................*
02C0 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 *................................*
02E0 00000000 00000000 00000000 00000000 4C4CE399 81838540 C5958440 40404C4C *................<<Trace End <<*
Figure 48. Sample CSUB trace — CICS connected to DB2 Version 5 or earlier
Figure 49 on page 194 shows the same situation when CICS is connected to DB2
Version 6 or later. In this example you can see that an identify, signon, and create
thread have been issued to DB2. There is an API request, followed by a syncpoint
and a dissociate (which dissociates the DB2 connection control block from the L8
TCB). The transaction now makes another API request, starting another unit of
work, and the DB2 connection control block is reassociated (ASSO) with the L8
TCB. A partial signon occurs to create an accounting record for the previous unit of
work. The API request is now issued to DB2.
Figure 49. Sample CSUB trace — CICS connected to DB2 Version 6 or later
In a CICS transaction dump, no summary or control blocks appear but the trace
table contains the CICS DB2 trace entries.
A sample showing CICS DB2 summary information from a CICS system dump is
shown in Figure 50 on page 195. It gives information on the global state of the
CICS DB2 connection, and a summary of transactions (under the headings ″Tran
id″, ″Task num″, ″TcaAddr″, ″TieAddr″, ″LotAddr″, ″Rctename″, ″RcteAddr″,
″CsubAddr″, ″Correlation id″, ″Uowid″, ″Subtask running″, and ″TCB in DB2″). This
sample was output when CICS was connected to DB2 Version 6, so using the open
transaction environment.
EDF screens for SQL statements are shown in Figure 51, and Figure 52 on page
197.
ENTER:
PF1 : UNDEFINED PF2 : UNDEFINED PF3 : UNDEFINED
PF4 : PF5 : PF6 :
PF7 : PF8 : PF9 :
PF10: PF11: UNDEFINED PF12:
ENTER: CONTINUE
PF1 : UNDEFINED PF2 : UNDEFINED PF3 : END EDF SESSION
PF4 : SUPPRESS DISPLAYS PF5 : WORKING STORAGE PF6 : USER DISPLAY
PF7 : SCROLL BACK PF8 : SCROLL FORWARD PF9 : STOP CONDITIONS
PF10: PREVIOUS DISPLAY PF11: UNDEFINED PF12: ABEND USER TASK
This section covers deadlocks only within DB2. If DB2 resources are involved in this
type of deadlock, one of the partners in the deadlock times out according to the
user-defined IRLM parameters. Other possible deadlocks are where resources
outside DB2 are involved.
Deadlocks are expected to occur, but not too often. You should give special
attention to deadlock situations if:
v Other transactions are often delayed because they access resources held by the
partners in the deadlock. This increases the response times for these
transactions. A cascade effect can then be the result.
v The resources involved in the deadlock are expected to be used more intensively
in the future, because of an increased transaction rate either for the transactions
involved in the deadlock or for other transactions.
The IRLM component of the DB2 subsystem performs deadlock detection at
user-defined intervals. One of the partners in the deadlock is the victim and
receives a -911 or a -913 return code from DB2. The actual return code is
determined by the DROLLBACK parameter for the DB2CONN (if a transaction is
using a pool thread) or the DB2ENTRY used by the transaction. The other partner
continues processing after the victim is rolled back.
Deadlock detection
In a normal production environment running without DB2 performance traces
activated, the easiest way to get information about a deadlock is to scan the MVS
log to find the messages shown in Figure 53.
From these messages, both partners in the deadlock are identified. The partners
Also, a second message identifies the resource that the victim could not obtain. The
other resource (whether it is the same or not) is not displayed in the message.
You should limit the DB2 performance trace to the two plans indicated in the MVS
log message. The "AUTH RCT" parameter specifies the CICS transaction ID; so
limiting the trace to the two transaction IDs (authorization IDs) involved can also be
reasonable. The performance trace to be started should include class(06) for
general locking events and class(03) for SQL events. The Database 2 Performance
Monitor (DB2PM) is a useful tool to format the trace output. The DB2PM lock
contention report and the lock suspension report can assist in determining the
resources involved in the deadlock.
If the output from the DB2PM reports is too large, you can develop a user program
to analyze the output from the traces. The goal is to find the resources involved in
the deadlock and all the SQL statements involved.
Making changes
In general, a deadlock occurs because two or more transactions both want the
same resources in opposite order at the same time and in a conflicting mode. The
actions taken to prevent a deadlock must deal with these characteristics.
Table 17 shows a list of preventive actions and the corresponding main effects.
Table 17. Deadlock prevention
Actions Spread Change the Decrease Change
Resources Locking Order Concurrency Locking Mode
Increase Index Freespace X
Increase Index Subpage Size X
Increase TS Freespace X
Change Clustering Index X X
Reorg the Table Space X X X
Add an Index X X (1)
Drop an Index X
To choose the right action, you must first understand why the deadlock occurred.
Then you can evaluate the actions to make your choices. These actions can have
several effects. They can:
v Solve the deadlock problem as desired
v Force a change in access path for other transactions causing new deadlocks
v Cause new deadlocks in the system.
It is therefore important that you carefully monitor the access path used by the
affected transactions, for example by the EXPLAIN facility in DB2. In many cases,
solving deadlocks is an iterative process.
PDF-only books
The following books are available in the CICS Information Center as Adobe
Portable Document Format (PDF) files:
Licensed publications
The following licensed publications are not included in the unlicensed version of the
Information Center:
CICS Diagnosis Reference, LY33-6102
CICS Data Areas, LY33-6103
CICS Supplementary Data Areas, LY33-6104
CICS Debugging Tools Interfaces Reference, LY33-6105
DB2
v DB2 Universal Database for OS/390 and z/OS Administration Guide, SC26-9931
v DB2 Universal Database for OS/390 and z/OS Application Programming and
SQL Guide, SC26-9933
v DB2 Universal Database for OS/390 and z/OS Command Reference, SC26-9934
v DB2 Universal Database for OS/390 and z/OS Data Sharing: Planning and
Administration, SC26-9935
v DB2 Universal Database Server for z/OS Data Sharing Quick Reference Card,
SX26-3846
v DB2 Universal Database for OS/390 and z/OS Diagnosis Guide and Reference,
LY37-3740
v DB2 Universal Database for OS/390 and z/OS Diagnostic Quick Reference,
LY37-3741
v DB2 Universal Database for OS/390 and z/OS Installation Guide, GC26-9936
v DB2 Universal Database for OS/390 and z/OS Messages and Codes,
GC26-9940
Bibliography 203
v DB2 Universal Database for OS/390 and z/OS Reference for Remote DRDA®
Requesters and Servers, SC26-9942
v DB2 Universal Database for OS/390 and z/OS Reference Summary, SX26-3847
v DB2 Universal Database for OS/390 and z/OS Release Planning Guide,
SC26-9943
v DB2 Universal Database for OS/390 and z/OS SQL Reference, SC26-9944
v DB2 Universal Database for OS/390 and z/OS Utility Guide and Reference,
SC26-9945
v DB2 Universal Database Server for OS/390 and z/OS What’s New?, GC26-9946
v DB2 Universal Database for OS/390 and z/OS Application Programming Guide
and Reference for Java, GC26-9932
v DB2 Universal Database for OS/390 and z/OS ODBC Guide and Reference,
GC26-9941
v DB2 Universal Database for OS/390 and z/OS Image, Audio, and Video
Extenders Administration and Programming, GC26-9947
v DB2 Universal Database for OS/390 and z/OS Net Search Extender
Administration and Programming, GC27-1171
v DB2 Universal Database for OS/390 and z/OS Text Extender Administration and
Programming, GC26-9941
v DB2 Universal Database for OS/390 and z/OS XML Administration and
Programming, GC26-9941
v An Introduction to DB2 Universal Database for z/OS, GC26-9937
v DB2 for OS/390 DB2 Packages: Implementation and Use, GG24-4001
Subsequent updates will probably be available in softcopy before they are available
in hardcopy. This means that at any time from the availability of a release, softcopy
versions should be regarded as the most up-to-date.
For CICS Transaction Server books, these softcopy updates appear regularly on the
Transaction Processing and Data Collection Kit CD-ROM, SK2T-0730-xx. Each
reissue of the collection kit is indicated by an updated order number suffix (the -xx
part). For example, collection kit SK2T-0730-06 is more up-to-date than
SK2T-0730-05. The collection kit is also clearly dated on the cover.
You can perform most tasks required to set up, run, and maintain your CICS system
in one of these ways:
v using a 3270 emulator logged on to CICS
v using a 3270 emulator logged on to TSO
v using a 3270 emulator as an MVS system console
IBM Personal Communications (Version 5.0.1 for Windows® 95, Windows 98,
Windows NT® and Windows 2000; version 4.3 for OS/2®) provides 3270 emulation
with accessibility features for people with disabilities. You can use this product to
provide the accessibility features you need in your CICS system.
G
GASET option 21 M
GETPAGE 159 macro changes 17
global trace records 147 MAXOPENTCBS system initialization parameter 52
GRANT command 83, 141 messages in problem determination 182
group attach 51 migrating to CICS DB2 attachment facility
and indoubt UOWs 27 assembling the RCT 17
and INITPARM system initialization parameter 52 RCTs to the CSD 23
CICS DB2 configuration requirements 15 to RDO 21
DB2GROUPID attribute 51 migration planning
identifying the chosen DB2 subsystem 52 DB2 databases 15
overriding with a specific DB2 subsystem 52 Mnotes 17
RESYNCMEMBER 27 MODIFY attachment facility command 43
GTF (generalized trace facility) 31, 146, 152 MODIFY TRACE command 146
monitoring data 145
monitoring DB2 152
H monitoring the attachment facility
handling deadlocks 197 CICS transactions 148, 151
held cursors 55, 113 functions 29
hot spots, examples 112 performance 148
tools 147
using CEMT commands 29
I using EXEC CICS commands 29
multithread connections 2
indoubt UOWs
resolution 26
INITPARM system initialization parameter
using 21
N
INQUIRE EXITPROGRAM 114 NOTCONNECTED 115
Installation and migration for CICS DB2 15 NUMLKTS 105
INVEXITREQ 114
IRLM component 197
isolation level 139 O
open TCBs
accounting 174
J application programs on 106
JCL requirements, CICS startup 15 as thread TCBs 6, 177
JDBC 117 open transaction environment (OTE)
acquiring a connection to a database 122 and application programs 106
Index 209
open transaction environment (OTE) (continued) RDO (resource definition online) (continued)
and enterprise beans 130 defining and installing DB2ENTRY 8
CICS DB2 configuration requirements 15 defining and installing DB2TRAN 8
CICS DB2 task-related user exit 106 RDONAME parameter 23
MAXOPENTCBS system initialization parameter RELEASE(COMMIT) 55
setting 52 RELEASE(DEALLOCATE) 62
processor times for transactions 174 releases of DB2 supported 15
TCBLIMIT setting 52 repeatable read 139
thread TCBs 6, 177 resynchronization information 28
threadsafe applications 106 RESYNCMEMBER 27
operations with CICS DB2 attachment facility RETAIN option 138
starting the CICS DB2 attachment facility 25 RETURN IMMEDIATE 114
stopping the CICS DB2 attachment facility 25 reusing threads
security 61
RMI (resource manager interface) 1
P RRCDTE sample job 68
package table (PT) 54
packages
advantages over dynamic plan switching 90 S
application design 90 sample connection exit routine (DSN3SATH) 77
converting existing applications 92 sample sign-on exit routine (DSN3SSGN) 81
overview 13 SASS (single address space) 76
page contention 112 security
performance AUTHTYPE 71
CICS DB2 attachment facility 148 command security 67
CICS transactions 151 DB2TRAN resource security 69
monitoring 147 defining RACF profiles 68
plans RACF 66
overview 13 RACF class, DSNR 76
pool threads 3, 59 resource security 68
problem determination SASS 76
CSUB trace 192 surrogate user checking 71
dump 194 security, surrogate 71
messages 182 sequential insertion 112
trace 182 sequential number allocation 112
wait types 177 serialization 111
processor time sign-on exit routine 81
calculating for DB2 Version 5 or earlier 174 single address space (SASS) 76
calculating for DB2 Version 6 or later 174 skeleton cursor table (SKCT) 54
class 1 167 skeleton package table (SKPT) 54
class 2 173 SMF (system management facility) 31, 145, 152
consumption 167 SMF 101 record
processor usage 167 fields 172
production procedure 140 special registers 55
programming features of JDBC and SQLJ 122 SQL
protected threads 57, 61 dynamic 85
PROTECTNUM 59, 61 qualified or unqualified 109
purging CICS DB2 transactions 30 static 85
SQL language 108
SQL processing, main activities 53
R SQL request 1
RACF 66 SQL return code
external security manager 66 -501 113
RACF class -818 138
DSNR 76 -911 197
RACF default resource profiles -913 197
VCICSCMD general resource class 70 SQLCA formatting routine 136
RACF list of groups option 81 SQLJ 117
RCT parameters. obsolete 17 acquiring a connection to a database 122
RDO (resource definition online) CICS abends 130
defining and installing DB2CONN 8 commit and rollback 129
Index 211
212 CICS TS for z/OS: CICS DB2 Guide
Notices
This information was developed for products and services offered in the U.S.A. IBM
may not offer the products, services, or features discussed in this document in other
countries. Consult your local IBM representative for information on the products and
services currently available in your area. Any reference to an IBM product, program,
or service is not intended to state or imply that only that IBM product, program, or
service may be used. Any functionally equivalent product, program, or service that
does not infringe any IBM intellectual property right may be used instead. However,
it is the user’s responsibility to evaluate and verify the operation of any non-IBM
product, program, or service.
IBM may have patents or pending patent applications covering subject matter
described in this document. The furnishing of this document does not give you any
license to these patents. You can send license inquiries, in writing, to:
For license inquiries regarding double-byte (DBCS) information, contact the IBM
Intellectual Property Department in your country or send inquiries, in writing, to:
The following paragraph does not apply in the United Kingdom or any other
country where such provisions are inconsistent with local law:
INTERNATIONAL BUSINESS MACHINES CORPORATION PROVIDES THIS
PUBLICATION “AS IS” WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESS
OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
OF NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A
PARTICULAR PURPOSE. Some states do not allow disclaimer of express or
implied warranties in certain transactions, therefore this statement may not apply to
you.
Licensees of this program who wish to have information about it for the purpose of
enabling: (i) the exchange of information between independently created programs
and other programs (including this one) and (ii) the mutual use of the information
which has been exchanged, should contact IBM United Kingdom Laboratories,
MP151, Hursley Park, Winchester, Hampshire, England, SO21 2JN. Such
information may be available, subject to appropriate terms and conditions, including
in some cases, payment of a fee.
Attention
Do not use this Diagnosis, Modification or Tuning Information as a
programming interface.
This book contains sample programs. Permission is hereby granted to copy and
store the sample programs into a data processing machine and to use the stored
copies for study and instruction only. No permission is granted to use the sample
programs for any other purpose.
Trademarks
The following terms are trademarks of International Business Machines Corporation
in the United States, or other countries, or both:
Java and all Java-based trademarks and logos are trademarks or registered
trademarks of Sun Microsystems Inc, in the United States, or other countries, or
both.
Other company, product, and service names may be trademarks or service marks
of others.
Notices 215
216 CICS TS for z/OS: CICS DB2 Guide
Sending your comments to IBM
If you especially like or dislike anything about this book, please use one of the
methods listed below to send your comments to IBM.
Feel free to comment on what you regard as specific errors or omissions, and on
the accuracy, organization, subject matter, or completeness of this book.
Please limit your comments to the information in this book and the way in which the
information is presented.
To ask questions, make comments about the functions of IBM products or systems,
or to request additional publications, contact your IBM representative or your IBM
authorized remarketer.
When you send comments to IBM, you grant IBM a nonexclusive right to use or
distribute your comments in any way it believes appropriate, without incurring any
obligation to you.
You can send your comments to IBM in any of the following ways:
v By mail, to this address:
User Technologies Department (MP095)
IBM United Kingdom Laboratories
Hursley Park
WINCHESTER,
Hampshire
SO21 2JN
United Kingdom
v By fax:
– From outside the U.K., after your international access code use
44–1962–816151
– From within the U.K., use 01962–816151
v Electronically, use the appropriate network ID:
– IBMLink™: HURSLEY(IDRCF)
– Internet: idrcf@hursley.ibm.com
Printed in USA
SC34-6252-00
Spine information:
Version 2
CICS TS for z/OS CICS DB2 Guide Release 3