DBMS Y3/S5

Concurrency Control

 Lock-Based Protocols
 Timestamp-Based Protocols
 Validation-Based Protocols
 Multiple Granularity
 Multiversion Schemes
 Deadlock Handling
 insert and Delete Operations and Predicate reads
 Weak Levels of Consistency
 Concurrency in Index Structure

Lock-Based Protocols

 A lock is a mechanism to control concurrent access to a data item

 Data items can be locked in two modes :
1. exclusive (X) mode. Data item can be both read as well as written. X-lock is requested
using lock-X instruction.

2. shared (S) mode. Data item can only be read. S-lock is requested using lock-S instruction.

 Lock requests are made to concurrency-control manager.Transaction can proceed only after
request is granted.
 Lock-compatibility matrix

 A transaction may be granted a lock on an item if the requested lock is compatible with locks
already held on the item by other transactions
 Any number of transactions can hold shared locks on an item,but if any transaction holds an
exclusive on the item no other transaction may hold any lock on the item.
 If a lock cannot be granted, the requesting transaction is made to wait till all incompatible locks
held by other transactions have been released. The lock is then granted.
 Example of a transaction performing locking:
T2: lock-S(A);

Dept of CSE ,RGCET Page 1

 DBMS Y3/S5
read (A);

unlock(A);

lock-S(B);

read (B);

unlock(B);

display(A+B)

 Locking as above is not sufficient to guarantee serializability — if A and Bget updated in-between
the read of A and B, the displayed sum would be wrong.
 A locking protocol is a set of rules followed by all transactions while requesting and releasing locks.
Locking protocols restrict the set of possible schedules.
Pitfalls of Lock-Based Protocols

 Consider the partial schedule

 Neither T3 nor T4 can make

progress — executing lock-S(B)
causes T4 to wait for T3 to release
its lock on B, while executing lock-
X(A) causesT3 to wait for T4 to release its lock on A.
 Such a situation is called a deadlock.
◦ To handle a deadlock one of T3 or T4 must be rolled back and its locks released.
 The potential for deadlock exists in most locking protocols.Deadlocks are a necessary evil.
 Starvation is also possible if concurrency control manager is badly designed. For example:
 A transaction may be waiting for an X-lock on an item, while a sequence of other transactions
request and are granted an S-lock on the same item.
 The same transaction is repeatedly rolled back due to deadlocks.
Concurrency control manager can be designed to prevent starvation.

The Two-Phase Locking Protocol

 This is a protocol which ensures conflict-serializable schedules.

 Phase 1: Growing Phase
◦ transaction may obtain locks
◦ transaction may not release locks
 Phase 2: Shrinking Phase
◦ transaction may release locks
◦ transaction may not obtain locks

Dept of CSE ,RGCET Page 2

 DBMS Y3/S5
 The protocol assures serializability. It can be proved that the transactions can be serialized in the
order of their lock points (i.e. the point where a transaction acquired its final lock).
 Two-phase locking does not ensure freedom from deadlocks
 Cascading roll-back is possible under two-phase locking. To avoid this, follow a modified protocol
called strict two-phase locking. Here a transaction must hold all its exclusive locks till it
commits/aborts.
 Rigorous two-phase locking is even stricter: here all locks are held till commit/abort. In this protocol
transactions can be serialized in the order in which they commit.
 There can be conflict serializable schedules that cannot be obtained if two-phase locking is
used.
 However, in the absence of extra information (e.g., ordering of access to data), two-phase locking is
needed for conflict serializability in the following sense:
Given a transaction Ti that does not follow two-phase locking, we can find a transaction Tj
that uses two-phase locking, and a schedule for Ti and Tj that is not conflict serializable.

Lock Conversion

 Two-phase locking with lock conversions:

– First Phase:

 can acquire a lock-S on item

 can acquire a lock-X on item
 can convert a lock-S to a lock-X (upgrade)
– Second Phase:

 can release a lock-S

 can release a lock-X
 can convert a lock-X to a lock-S (downgrade)
This protocol assures serializability. But still relies on the programmer to insert the various locking
instructions.

Automatic Acquisition of Locks

 A transaction Ti issues the standard read/write instruction,

without explicit locking calls.

 The operation read(D) is processed as:

if Ti has a lock on D

then

read(D)

else

begin
Dept of CSE ,RGCET Page 3
 DBMS Y3/S5
if necessary wait until no other

transaction has a lock-X on D

grant Ti a lock-S on D;

read(D)

end

 write(D) is processed as:

 if Ti has a lock-X on D
then

write(D)

else

begin

if necessary wait until no other trans. has any lock on D,

if Ti has a lock-S on D

then

upgrade lock on D to lock-X

else

grant Ti a lock-X on D

write(D)

end;

 All locks are released after commit or abort

Implementation of Locking

 A Lock manager can be implemented as a separate process to which transactions send lock and
unlock requests
 The lock manager replies to a lock request by sending a lock grant messages (or a message asking
the transaction to roll back, in case of a deadlock)
 The requesting transaction waits until its request is answered
 The lock manager maintains a datastructure called a lock table to record granted locks and pending
requests
 The lock table is usually implemented as an in-memory hashtable indexed on the name of the data
item being locked

Dept of CSE ,RGCET Page 4

 DBMS Y3/S5

Lock tables

 Black rectangles indicate grantedlocks, white ones indicate waiting requests

 Lock table also records the type of lock granted or requested
 New request is added to the end of the queue of requests for the data item, and granted if it is
compatible with all earlier locks
 Unlock requests result in the request being deleted, and later requests are checked to see if they
can now be granted
 If transaction aborts, all waiting or granted requests of the transaction are deleted lock manager
may keep a list of locks held by each transaction, to Implement this efficiently
Graph-Based Protocols

 Graph-based protocols are an alternative to two-phase locking

◦ Impose a partial ordering → on the set D = {d1, d2 ,..., dh} of all
data items.

 If di → dj then any transaction accessing both di and dj must access

di before accessing dj.

 Implies that the set D may now be viewed as a directed acyclic

graph, called a database graph.

 The tree-protocol is a simple kind of graph protocol.

Tree Protocol

Dept of CSE ,RGCET Page 5

 DBMS Y3/S5

 Only exclusive locks are allowed.

 The first lock by Ti may be on any data item. Subsequently, a data Q can be locked by Ti only if the
parent of Q is currently locked by Ti.
 Data items may be unlocked at any time.
 The tree protocol ensures conflict serializability as well as freedom from deadlock.
 Unlocking may occur earlier in the tree-locking protocol than in the two-phase locking protocol.
 shorter waiting times, and increase in concurrency
 protocol is deadlock-free, no rollbacks are required
 the abort of a transaction can still lead to cascading rollbacks.
(this correction has to be made in the book also.)

 However, in the tree-locking protocol, a transaction may have to

lock data items that it does not access.

 increased locking overhead, and additional waiting time

 potential decrease in concurrency
Schedules not possible under two-phase locking are possible under tree protocol, and vice versa.

Timestamp-Based Protocols

 Each transaction is issued a timestamp when it enters the system. If an old transaction Ti has time-
stamp TS(Ti), a new transaction Tj is assigned time-stamp TS(Tj) such that TS(Ti) <TS(Tj).
 The protocol manages concurrent execution such that the time- stamps determine the
serializability order.

Dept of CSE ,RGCET Page 6

 DBMS Y3/S5
 In order to assure such behavior, the protocol maintains for each data Q two timestamp values:
 W-timestamp(Q) is the largest time-stamp of any transaction that executed write(Q)
successfully.
 R-timestamp(Q) is the largest time-stamp of any transaction that executed read(Q) successfully.
 The timestamp ordering protocol ensures that any conflicting read and write operations are
executed in timestamp order.
 Suppose a transaction Ti issues a read(Q)
1. If TS(Ti) ≤ W-timestamp(Q), then Ti needs to read a value of Q that was already overwritten.
Hence, the read operation is rejected, and Ti is rolled back.

2. If TS(Ti)≥ W-timestamp(Q), then the read operation is executed, and R-timestamp(Q) is set to the
maximum of R-

timestamp(Q) and TS(Ti).

 Suppose that transaction Ti issues write(Q).

 If TS(Ti) < R-timestamp(Q), then the value of Q that Ti is producing was needed previously, and the
system assumed that value would never be produced. Hence, the write operation is rejected, and Ti
is rolled back.
 If TS(Ti) < W-timestamp(Q), then Ti is attempting to write an obsolete value of Q. Hence, this write
operation is rejected, and Ti is rolled back.
 Otherwise, the write operation is executed, and W- timestamp(Q) is set to TS(Ti).

Correctness of Timestamp-Ordering Protocol

 The timestamp-ordering protocol guarantees serializability since all the arcs in the precedence
graph are of the form:
Thus, there will be no cycles in the precedence graph

 Timestamp protocol ensures freedom from deadlock as no transaction ever waits.

 But the schedule may not be cascade-free, and may not even be recoverable.
Recoverability and Cascade Freedom

 Problem with timestamp-ordering protocol:

◦ Suppose Ti aborts, but Tj has read a data item written by Ti
◦ Then Tj must abort; if Tj had been allowed to commit earlier, the schedule is not recoverable.
 Further, any transaction that has read a data item written by Tj must abort
 This can lead to cascading rollback --- that is, a chain of rollbacks

Solution:

 A transaction is structured such that its writes are all performed at the end of its processing
 All writes of a transaction form an atomic action; no transaction may execute while a transaction is
being written
 A transaction that aborts is restarted with a new timestamp

Dept of CSE ,RGCET Page 7

 DBMS Y3/S5

Thomas’ Write Rule

 Modified version of the timestamp-ordering protocol in which obsolete write operations may be
ignored under certain circumstances.
 When Ti attempts to write data item Q, if TS(Ti) < W-timestamp(Q), then Ti is attempting to write
an obsolete value of{Q}. Hence, rather than rolling back Ti as the timestamp ordering
protocol would have done, this {write} operation can be ignored.

 Otherwise this protocol is the same as the timestamp ordering protocol.

 Thomas' Write Rule allows greater potential concurrency. Unlike previous protocols, it
allows some view-serializable schedules that are not conflict-serializable.

Validation-Based Protocol

 Execution of transaction Ti is done in three phases.

1. Read and execution phase: Transaction Ti writes only to temporary local variables

2. Validation phase: Transaction Ti performs a ``validation test'' to determine if local variables can
be written without violating serializability.

3. Write phase: If Ti is validated, the updates are applied to the database; otherwise, Ti is rolled
back.

 The three phases of concurrently executing transactions can be interleaved, but each transaction
must go through the three phases in that order.
 Also called as optimistic concurrency control since transaction executes fully in the hope that all
will go well during validation
 Each transaction Ti has 3 timestamps
− Start(Ti) : the time when Ti started its execution

− Validation(Ti): the time when Ti entered its validation phase

− Finish(Ti) : the time when Ti finished its write phase

 Serializability order is determined by timestamp given at validation time, to increase

concurrency. Thus TS(Ti) is given the value of Validation(Ti).
 This protocol is useful and gives greater degree of concurrency if probability of conflicts is
low. That is because the serializability order is not pre-decided and relatively less transactions will
have to be rolled back.

Validation Test for Transaction Tj

Dept of CSE ,RGCET Page 8

 DBMS Y3/S5
 If for all Ti with TS (Ti) < TS (Tj) either one of the following condition holds:
 finish(Ti) < start(Tj)
 start(Tj) < finish(Ti) < validation(Tj) and the set of data items written by Ti does not intersect with
the set of data items read by Tj. then validation succeeds and Tj can be committed.
Otherwise,validation fails and Tj is aborted.
 Justification: Either first condition is satisfied, and there is no overlapped execution, or second
condition is satisfied and
1. the writes of Tj do not affect reads of Ti since they occur after Ti has finished its reads.

2. the writes of Ti do not affect reads of Tj since Tj does not read any item written by Ti.

Schedule Produced by Validation

! Example of schedule produced using validation

T14 T15

read(B)

read(B)

B:- B-50

read(A)

A:- A+50

read(A)

(validate)

display (A+B)

(validate)

write (B)

write (A)

Multiple Granularity

 Allow data items to be of various sizes and define a hierarchy of data granularities, where the small
granularities are nested within larger ones
 Can be represented graphically as a tree (but don't confuse with tree-locking protocol)
 When a transaction locks a node in the tree explicitly, it implicitly locks all the node's

Dept of CSE ,RGCET Page 9

 DBMS Y3/S5
descendents in the same mode.
 Granularity of locking (level in tree where locking is done):
◦ fine granularity (lower in tree): high concurrency, high locking overhead
 coarse granularity (higher in tree): low locking overhead, low concurrency
Intention Lock Modes

 In addition to S and X lock modes, there are three additional lock modes with multiple
granularity:
 intention-shared (IS): indicates explicit locking at a lower level of the tree but only with shared
locks.
 intention-exclusive (IX): indicates explicit locking at a lower level with exclusive or shared locks
 shared and intention-exclusive (SIX): the subtree rooted by that node is locked explicitly in
shared mode and explicit locking is being done at a lower level with exclusive-mode locks.
 intention locks allow a higher level node to be locked in S or Xmode without having to check all
descendent nodes.

Compatibility Matrix with Intention Lock Modes

! The compatibility matrix for all lock modes is:

IS IX S SIX X

IS " "" " ×

IX " "×××

S "×"××

SIX "××××

X ×××××

Dept of CSE ,RGCET Page 10

 DBMS Y3/S5

Multiple Granularity Locking Scheme

 Transaction Ti can lock a node Q, using the following rules:

1. The lock compatibility matrix must be observed.

2. The root of the tree must be locked first, and may be locked in any mode.

3. A node Q can be locked by Ti in S or IS mode only if the parent of Q is currently locked by Ti in

either IX or IS mode.

4. A node Q can be locked by Ti in X, SIX, or IX mode only if the parent of Q is currently locked by Ti
in either IX or SIX mode.

5. Ti can lock a node only if it has not previously unlocked any node (that is, Ti is two-phase).

6. Ti can unlock a node Q only if none of the children of Q are currently locked by Ti.

 Observe that locks are acquired in root-to-leaf order,whereas they are released in leaf-to-root
order.

Multiversion Schemes

 Multiversion schemes keep old versions of data item to increase concurrency.

 Multiversion Timestamp Ordering
 Multiversion Two-Phase Locking
Each successful write results in the creation of a new version of the data item written.

 Use timestamps to label versions.

 When a read(Q) operation is issued, select an appropriate version of Q based on the timestamp of
the transaction, and return the value of the selected version.
 reads never have to wait as an appropriate version is returned immediately.

Multiversion Timestamp Ordering

 Each data item Q has a sequence of versions <Q1, Q2,...., Qm>.

Each version Qk contains three data fields:

 Content -- the value of version Qk.

 W-timestamp(Qk) -- timestamp of the transaction that created (wrote) version Qk
 R-timestamp(Qk) -- largest timestamp of a transaction thatsuccessfully read version Qk
 when a transaction Ti creates a new version Qk of Q, Qk's W-timestamp and R-timestamp are
initialized to TS(Ti).
 R-timestamp of Qk is updated whenever a transaction Tj readsQk, and TS(Tj) > R-timestamp(Qk).
 The multiversion timestamp scheme presented next ensures serializability.
Suppose that transaction Ti issues a read(Q) or write(Q) operation.

Dept of CSE ,RGCET Page 11

 DBMS Y3/S5
Let Qk denote the version of Q whose write timestamp is the largest write timestamp less than or
equal to TS(Ti).

1. If transaction Ti issues a read(Q), then the value returned is the content of version Qk.

2. If transaction Ti issues a write(Q), and if TS(Ti) < R-timestamp(Qk), then transaction Ti is rolled
back. Otherwise, if TS(Ti) = W-timestamp(Qk), the contents of Qk are overwritten, otherwise a new
version of Q is created.

Reads always succeed; a write by Ti is rejected if some other transaction Tj that (in the serialization
order defined by the timestamp values) should read Ti's write, has already read a version created
by a transaction older than Ti.

Multiversion Two-Phase Locking

 Differentiates between read-only transactions and update transactions

 Update transactions acquire read and write locks, and hold all locks up to the end of the
transaction. That is, update transactions follow rigorous two-phase locking.
 Each successful write results in the creation of a new version of the data item written.
each version of a data item has a single timestamp whose value is obtained from a counter
ts-counter that is incremented during commit processing.

Read-only transactions are assigned a timestamp by reading the current value of ts-counter before
they start execution; they follow the multiversion timestamp-ordering protocol for

performing reads.

 When an update transaction wants to read a data item, it obtains a shared lock on it, and reads the
latest version.
 When it wants to write an item, it obtains X lock on; it then creates a new version of the item and
sets this version's timestamp to ∞.
 When update transaction Ti completes, commit processing
occurs:

Ti sets timestamp on the versions it has created to ts-counter + 1

Ti increments ts-counter by 1

Read-only transactions that start after Ti increments ts-counter will see the values updated by Ti.

Read-only transactions that start before Ti increments the ts-counter will see the value before the
updates by Ti.Only serializable schedules are produced.

Dept of CSE ,RGCET Page 12

 DBMS Y3/S5
Deadlock Handling

! Consider the following two transactions:

T1:

write (X)

write(Y)

T2:

write(Y)

write(X)

 Schedule with deadlock

T1 T2

lock-X on X

write (X)

lock-X on Y

write (X)

wait for lock-X on X

wait for lock-X on Y

Deadlock Handling

 System is deadlocked if there is a set of transactions such that every transaction in the set is waiting
for another transaction in the set.
 Deadlock prevention protocols ensure that the system will never enter into a deadlock state.
Some prevention strategies :
 Require that each transaction locks all its data items before it begins execution (predeclaration).
 Impose partial ordering of all data items and require that a transaction can lock data items
only in the order specified by the partial order (graph-based protocol).

More Deadlock Prevention Strategies

Dept of CSE ,RGCET Page 13

 DBMS Y3/S5
 Following schemes use transaction timestamps for the sake of deadlock prevention alone.
 wait-die scheme — non-preemptive
◦ older transaction may wait for younger one to release data item.
Younger transactions never wait for older ones; they are rolled back instead.

 a transaction may die several times before acquiring needed data item
 wound-wait scheme — preemptive
◦ older transaction wounds (forces rollback) of younger transaction instead of waiting for it.
Younger transactions may wait for older ones.
 may be fewer rollbacks than wait-die scheme.

 Both in wait-die and in wound-wait schemes, a rolled back transactions is restarted with its original
timestamp. Older transactions thus have precedence over newer ones, and starvation is hence
avoided.

 Timeout-Based Schemes :
◦ a transaction waits for a lock only for a specified amount of time.
After that, the wait times out and the transaction is rolled back.

 thus deadlocks are not possible

 simple to implement; but starvation is possible. Also difficult to determine good value of the
timeout interval.
Deadlock Detection

 Deadlocks can be described as a wait-for graph, which consists of a pair G = (V,E),

 V is a set of vertices (all the transactions in the system)
E is a set of edges; each element is an ordered pair Ti →Tj.

 If Ti → Tj is in E, then there is a directed edge from Ti to Tj,implying that Ti is waiting for Tj to

release a data item.
 When Ti requests a data item currently being held by Tj, then the edge Ti Tj is inserted in the
wait-for graph. This edge is removed only when Tj is no longer holding a data item needed by Ti.
 The system is in a deadlock state if and only if the wait-for graphhas a cycle. Must invoke a
deadlock-detection algorithm periodically to look for cycles.
Wait-for graph without a cycle

Wait-for graph with a cycle

Deadlock Recovery

 When deadlock is detected :

 Some transaction will have to rolled back (made a victim) to break deadlock. Select that transaction
as victim that will incur minimum cost.
 Rollback -- determine how far to roll back transaction
# Total rollback: Abort the transaction and then restart it.

# More effective to roll back transaction only as far as necessary to break deadlock.

Dept of CSE ,RGCET Page 14

 DBMS Y3/S5
 Starvation happens if same transaction is always chosen as victim.
Include the number of rollbacks in the cost factor to avoid starvation.

Insert Operations, Delete Operations, and Predicate Reads

 If two-phase locking is used :

• delete(Q) deletes data item Q from the database.

• insert(Q) inserts a new data item Q into the database and assigns Q an initial value.

 A delete operation may be performed only if the transaction deleting the tuple has an exclusive
lock on the tuple to be deleted.
 A transaction that inserts a new tuple into the database is given an X-mode lock on the tuple
 Insertions and deletions can lead to the phantom phenomenon.
 A transaction that scans a relation (e.g., find all accounts in Perryridge) and a transaction that
inserts a tuple in the relation (e.g.,insert a new account at Perryridge) may conflict in spite of not
accessing any tuple in common.
 If only tuple locks are used, non-serializable schedules can result: the scan transaction may
not see the new account, yet may be serialized before the insert transaction.
 The transaction scanning the relation is reading information that indicates what tuples the relation
contains, while a transaction inserting a tuple updates the same information.
 The information should be locked.
 One solution:
 Associate a data item with the relation, to represent the information about what tuples the relation
contains.
 Transactions scanning the relation acquire a shared lock in the data item,
 Transactions inserting or deleting a tuple acquire an exclusive lock on the data item. Above
protocol provides very low concurrency for insertions/deletions.
 Index locking protocols provide higher concurrency while preventing the phantom
phenomenon, by requiring locks on certain index buckets.
Deletion

Let Ii and I j be instructions of Ti and Tj , respectively, that appear in schedule S in consecutive order. Let Ii =
delete(Q).We consider several instructions I j .

• I j = read(Q). Ii and I j conflict. If Ii comes before I j , Tj will have a logical error. If I j comes before Ii , Tj can
execute the read operation successfully.

• I j = write(Q). Ii and I j conflict. If Ii comes before I j , Tj will have a logical error. If I j comes before Ii , Tj can
execute the write operation successfully.

• I j = delete(Q). Ii and I j conflict. If Ii comes before I j , Ti will have a logical error. If I j comes before Ii , Ti
will have a logical error.

Dept of CSE ,RGCET Page 15

 DBMS Y3/S5
• I j = insert(Q). Ii and I j conflict. Suppose that data item Q did not exist prior to the execution of Ii and I j .
Then, if Ii comes before I j , a logical error results for Ti. If I j comes before Ii , then no logical error results.
Likewise, if Q existed prior to the execution of Ii and I j , then a logical error results if I j comes before Ii, but
not otherwise.

We can conclude the following:

• Under the two-phase locking protocol, an exclusive lock is required on a data item before that item can
be deleted.

• Under the timestamp-ordering protocol, a test similar to that for a write must be performed. Suppose
that transaction Ti issues delete(Q).

◦ If TS(Ti ) < R-timestamp(Q), then the value of Q that Ti was to delete has already been read by a
transaction Tj with TS(Tj ) > TS(Ti ). Hence, the delete operation is rejected, and Ti is rolled back.

◦ If TS(Ti )< W-timestamp(Q), then a transaction Tj with TS(Tj )> TS(Ti) has written Q. Hence, this delete
operation is rejected, and Ti is rolled back.

◦ Otherwise, the delete is executed.

Insertion

insert(Q) conflictswitharead(Q) operation or a write(Q) operation; no read or write can be performed on a

data item before it exists.

Since an insert(Q) assigns a value to data item Q, an insert is treated similarly to a write for concurrency-
control purposes:

• Under the two-phase locking protocol, if Ti performs an insert(Q) operation,Ti is given an exclusive lock
on the newly created data item Q.

• Under the timestamp-ordering protocol, if Ti performs an insert(Q) operation,the values R-timestamp(Q)

andW-timestamp(Q) are set to TS(Ti ).

Predicate Reads and The Phantom Phenomenon

Consider transaction T30 that executes the following SQL query on the university database:

select count(*)

from instructor

where dept name = ’Physics’ ;

Dept of CSE ,RGCET Page 16
 DBMS Y3/S5

Transaction T30 requires access to all tuples of the instructor relation pertaining to the Physics department.

Let T31 be a transaction that executes the following SQL insertion:

insert into instructor

values (11111,’Feynman’, ’Physics’, 94000);

Let S be a schedule involving T30 and T31. We expect there to be potential for

a conflict for the following reasons:

• If T30 uses the tuple newly inserted by T31 in computing count(*), then T30 reads a value written by T31.
Thus, in a serial schedule equivalent to S, T31 must come before T30.

• If T30 does not use the tuple newly inserted by T31 in computing count(*), then in a serial schedule
equivalent to S, T30 must come before T31.

The second of these two cases is curious. T30 and T31 do not access any tuple in common, yet they conflict
with each other! In effect, T30 and T31 conflict on a phantom tuple. If concurrency control is performed at
the tuple granularity, this conflict would go undetected. As a result, the system could fail to prevent a
nonserializable schedule. This problem is called the phantom phenomenon.

Index Locking Protocol

 Every relation must have at least one index. Access to a relation must be made only through one of
the indices on the relation.
 A transaction Ti that performs a lookup must lock all the index buckets that it accesses, in S-mode.
 A transaction Ti may not insert a tuple ti into a relation r without updating all indices to r.
 Ti must perform a lookup on every index to find all index buckets that could have possibly
contained a pointer to tuple ti, had it existed already, and obtain locks in X-mode on all these
index buckets. Ti must also obtain locks in X-mode on all index buckets
that it modifies.

 The rules of the two-phase locking protocol must be observed.

Dept of CSE ,RGCET Page 17

 DBMS Y3/S5
Weak Levels of Consistency

 Degree-two consistency: differs from two-phase locking in that S-locks may be released at any
time, and locks may be acquired at any time
 X-locks must be held till end of transaction
 Serializability is not guaranteed, programmer must ensure that no erroneous database state will
occur.

Fig: Nonserializable schedule with degree-two consistency.

Cursor stability:

Cursor stability is a form of degree-two consistency designed for programs that iterate over tuples of a
relation by using cursors. Instead of locking the entire relation, cursor stability ensures that:

• The tuple that is currently being processed by the iteration is locked in shared mode.

• Any modified tuples are locked in exclusive mode until the transaction commits.

These rules ensure that degree-two consistency is obtained.

 Two-phase locking is not required.

 Serializability is not guaranteed.
Cursor stability is used in practice on heavily accessed relations as a means of increasing concurrency and
improving system performance. Applications that use cursor stability must be coded in a way that ensures
database consistency despite the possibility of non serializable schedules. Thus, the use of cursor stability is
limited to specialized situations with simple consistency constraints.

Concurrency Control Across User Interactions

Dept of CSE ,RGCET Page 18

 DBMS Y3/S5

Concurrency-control protocols usually consider transactions that do not involve user interaction.

• For each updated tuple, the transaction checks if the current version number is the same as the version
number of the tuple when it was first read by the transaction.

1. If the version numbers match, the update is performed on the tuple in the database, and its version
number is incremented by 1.

2. If the version numbers do not match, the transaction is aborted, rolling back all the updates it
performed.

Transactions that involve user interaction are called conversations in Hibernate to differentiate them from
regular transactions validation using version numbers is very useful for such transactions.

SNAPSHOT ISOLATION

Snapshot isolation is a particular type of concurrency-control scheme that has gained wide
acceptance in commercial and open-source systems, including Oracle, PostgreSQL, and SQL Server.

Validation Steps for Update Transactions

Deciding whether or not to allow an update transaction to commit requires some care. Potentially,
two transactions running concurrently might both update the same data item. Since these two transactions
operate in isolation using their own private snapshots, neither transaction sees the update made by the
other. If both transactions are allowed to write to the database, the first update written will be overwritten
by the second. The result is a lost update. Clearly, this must be prevented. There are two variants of
snapshot isolation, both of which prevent lost updates. They are called first committerwins and first
updaterwins. Both approaches are based on testing the transaction against concurrent
transactions.Atransaction is said to be concurrent with T if it was active or partially committed at any point
from the start of T up to and including the timewhen this test is being performed.

Under first committer wins, when a transaction T enters the partially committed state, the
following actions are taken in an atomic action:

A test is made to see if any transaction that was concurrentwith T has already written an update to the
database for some data item that T intends to write.

• If some such transaction is found, then T aborts.

• If no such transaction is found, then T commits and its updates are written to the database.

This approach is called “first committer wins” because if transactions conflict, the first one to be tested
using the above rule succeeds in writing its updates, while the subsequent ones are forced to abort.

Under first updater wins the system uses a locking mechanism that applies only to updates.
Dept of CSE ,RGCET Page 19
 DBMS Y3/S5
When a transaction Ti attempts to update a data item, it requests a write lock on that data item. If the lock
is not held by a concurrent transaction, the following steps are taken after the lock is acquired:

• If the item has been updated by any concurrent transaction, then Ti aborts.

• Otherwise Ti may proceed with its execution including possibly committing.

If, however, some other concurrent transaction Tj already holds a write lock on that data item, then Ti
cannot proceed and the following rules are followed:

• Ti waits until Tj aborts or commits.

◦ If Tj aborts, then the lock is released and Ti can obtain the lock. After the lock is acquired, the check for an
update by a concurrent transaction is performed as described earlier: Ti aborts if a concurrent transaction
had updated the data item, and proceeds with its execution otherwise.

◦ If Tj commits, then Ti must abort.

Locks are released when the transaction commits or aborts.

This approach is called “first updater wins” because if transactions conflict, the first one to obtain the lock is
the one that is permitted to commit and perform its update. Those that attempt the update later abort
unless the first updater subsequently aborts for some other reason.

Serializability Issues

Snapshot isolation is attractive in practice because the overhead is low and no aborts occur unless two
concurrent transactions update the same data item.

There is, however, one serious problemwith the snapshot isolation scheme as we have presented it, and as
it is implemented in practice: snapshot isolation does not ensure serializability.

1. Suppose that we have two concurrent transactions Ti and Tj and two data items A and B. Suppose that Ti
reads A and B, then updates B, while Tj reads Aand B, then updates A. For simplicity, we assume there are
no other concurrent transactions. Since Ti and Tj are concurrent, neither transaction

sees the update by the other in its snapshot. But, since they update different data items, both are allowed
to commit regardless of whether the system uses the first-update-wins policy or the first-committer-wins
policy.

However, the precedence graph has a cycle. There is an edge in the precedence graph from Ti to Tj because
Ti reads the value of Athat existed before Tj writes A. There is also an edge in the precedence graph from Tj
to Ti because Tj reads the value of B that existed before Ti writes B. Since there is a cycle in the precedence
graph, the result is a nonserializable schedule.

This situation, where each of a pair of transactions has read data that is written by the other, but there is
no data written by both transactions, is referred to as write skew.
Dept of CSE ,RGCET Page 20
 DBMS Y3/S5
Chapter 4.3 : Recovery System

 Failure Classification
 Storage Structure
 Recovery and Atomicity
 Log-Based Recovery
 Shadow Paging
 Recovery With Concurrent Transactions
 Buffer Management
 Failure with Loss of Nonvolatile Storage
 Early Lock Release and Logical Undo Operations

Failure Classification

 Transaction failure :
 Logical errors: transaction cannot complete due to some internal error condition
 System errors: the database system must terminate an active transaction due to an error
condition (e.g., deadlock)
 System crash: a power failure or other hardware or software failure causes the system to crash.
 Fail-stop assumption: non-volatile storage contents are assumed to not be corrupted by system
crash Database systems have numerous integrity checks to prevent corruption of disk data
 Disk failure: a head crash or similar disk failure destroys all or part of disk storage
 Destruction is assumed to be detectable: disk drives use checksums to detect failures

Storage

Volatile storage:

 does not survive system crashes

 examples: main memory, cache memory

Nonvolatile storage:

 survives system crashes

 examples: disk, tape, flash memory,
 non-volatile (battery backed up) RAM

Stable storage:

Dept of CSE ,RGCET Page 21

 DBMS Y3/S5

 a mythical form of storage that survives all failures

 approximated by maintaining multiple copies on distinct nonvolatile
media

Stable-Storage Implementation

 Maintain multiple copies of each block on separate disks

◦ copies can be at remote sites to protect against disasters such as fire or flooding.
 Failure during data transfer can still result in inconsistent copies: Block transfer can result in
◦ Successful completion
◦ Partial failure: destination block has incorrect information
◦ Total failure: destination block was never updated
 Protecting storage media from failure during data transfer (one solution):
◦ Execute output operation as follows (assuming two copies of each block):
1. Write the information onto the first physical block.

2. When the first write successfully completes, write the same information onto the second
physical block.

3. The output is completed only after the second write successfully completes.

 Copies of a block may differ due to failure during output operation. To recover from failure:
1. First find inconsistent blocks:

1. Expensive solution: Compare the two copies of every disk block.

2. Better solution:

 Record in-progress disk writes on non-volatile storage (Non-volatile RAM or special

area of disk).
 Use this information during recovery to find blocks that may be
 inconsistent, and only compare copies of these.
 Used in hardware RAID systems
2. If either copy of an inconsistent block is detected to have an error (bad checksum), overwrite it
by the other copy. If both have no error, but are different, overwrite the second block by the first block.

Data Access

 Physical blocks are those blocks residing on the disk.

 Buffer blocks are the blocks residing temporarily in main memory.

Dept of CSE ,RGCET Page 22

 DBMS Y3/S5
 Block movements between disk and main memory are initiated through the following two
operations:
◦ input(B) transfers the physical block B to main memory.
◦ output(B) transfers the buffer block B to the disk, and replaces theappropriate physical block
there.
 Each transaction Ti has its private work-area in which local copies of all data items accessed and
updated by it are kept.
 Ti's local copy of a data item X is called xi.

We assume, for simplicity, that each data item fits in, and is stored inside, a single block.

 Transaction transfers data items between system buffer blocks and its private work-area using the
following operations :
 read(X) assigns the value of data item X to the local variable xi.
 write(X) assigns the value of local variable xi to data item {X} in the buffer block.
 both these commands may necessitate the issue of an input(BX) instruction before the
assignment, if the block BX in which X resides is not already in memory.
 Transactions
◦ Perform read(X) while accessing X for the first time;
◦ All subsequent accesses are to the local copy.
◦ After last access, transaction executes write(X).
 output(BX) need not immediately follow write(X). System can perform the output operation when
it deems fit.

Recovery and Atomicity

 Modifying the database without ensuring that the transaction will commit may leave the database
in an inconsistent state.
 Consider transaction Ti that transfers $50 from account A to account B; goal is either to perform all
database modifications made by Ti or none at all.
 Several output operations may be required for Ti (to output A and B). A failure may occur after one
of these modifications have been made but before all of them are made.
 To ensure atomicity despite failures, we first output information describing the modifications to
stable storage without modifying the database itself.
 We study two approaches:
◦ log-based recovery, and
◦ shadow-paging
 We assume (initially) that transactions run serially, that is, one after the other.

Log-Based Recovery

 A log is kept on stable storage.

Dept of CSE ,RGCET Page 23

 DBMS Y3/S5
◦ The log is a sequence of log records, and maintains a record of update activities on the
database.
 When transaction Ti starts, it registers itself by writing a <Ti start>log record
 Before Ti executes write(X), a log record <Ti, X, V1, V2> is written, where V1 is the value of X before
the write, and V2 is the value to be written to X.
 Log record notes that Ti has performed a write on data item Xj Xj had value V1 before the write,
and will have value V2 after the write.
 When Ti finishes it last statement, the log record <Ti commit> is written.
 We assume for now that log records are written directly to stable storage (that is, they are not
buffered)
 Two approaches using logs
◦ Deferred database modification
◦ Immediate database modification
Deferred Database Modification

The deferred database modification scheme records all modifications to the log, but defers all the writes to
after partial commit.

 Assume that transactions execute serially

 Transaction starts by writing <Ti start> record to log. A write(X) operation results in a log record <Ti,
X, V> being written, where V is the new value for X
Note: old value is not needed for this scheme

 The write is not performed on X at this time, but is deferred. When Ti partially commits, <Ti
commit> is written to the log Finally, the log records are read and used to actually execute the
previously deferred writes.
 During recovery after a crash, a transaction needs to be redone if and only if both <Ti start> and<Ti
commit> are there in the log.
 Redoing a transaction Ti ( redoTi) sets the value of all data items updated by the transaction to the
new values.
 Crashes can occur while the transaction is executing the original updates, or while recovery action
is being taken example transactions T0 and T1 (T0 executes before T1):
T0: read (A)

A: - A - 50

Write (A)

read (B)

B:- B + 50

write (B)

T1 : read (C)

Dept of CSE ,RGCET Page 24

 DBMS Y3/S5
C:- C- 100

write ©

 Below we show the log as it appears at three instances of time.

 If log on stable storage at time of crash is as in case:

(a) No redo actions need to be taken

(b) redo(T0) must be performed since <T0 commit> is present

(c) redo(T0) must be performed followed by redo(T1) since

<T0 commit> and <Ti commit> are present

Immediate Database Modification

Dept of CSE ,RGCET Page 25

 DBMS Y3/S5

 The immediate database modification scheme allows database updates of an uncommitted

transaction to be made as the writes are issued
 since undoing may be needed, update logs must have both old value and new value
 Update log record must be written before database item is written
◦ We assume that the log record is output directly to stable storage
◦ Can be extended to postpone log record output, so long as prior to execution of an output(B)
operation for a data block B, all log records corresponding to items B must be flushed to stable
storage
 Output of updated blocks can take place at any time before or after transaction commit
 Order in which blocks are output can be different from the order in which they are written.
 Recovery procedure has two operations instead of one:
◦ undo(Ti) restores the value of all data items updated by Ti to their old values, going backwards
from the last log record for Ti
 redo(Ti) sets the value of all data items updated by Ti to the new values, going forward from
the first log record for Ti
 Both operations must be idempotent
◦ That is, even if the operation is executed multiple times the effect is the same as if it is
executed once Needed since operations may get re-executed during recovery
 When recovering after failure:
◦ Transaction Ti needs to be undone if the log contains the record
<Ti start>, but does not contain the record <Ti commit>.

 Transaction Ti needs to be redone if the log contains both the record

<Ti start> and the record <Ti commit>.

 Undo operations are performed first, then redo operations.

Immediate DB Modification Recovery

Example

Below we show the log as it appears at three instances of time.

Dept of CSE ,RGCET Page 26

 DBMS Y3/S5
Recovery actions in each case above are:

(a) undo (T0): B is restored to 2000 and A to 1000.

(b) undo (T1) and redo (T0): C is restored to 700, and then A and B are set to 950 and 2050
respectively.

(c) redo (T0) and redo (T1): A and B are set to 950 and 2050 respectively. Then C is set to 600

Checkpoints

 Problems in recovery procedure as discussed earlier :

1. searching the entire log is time-consuming

2. we might unnecessarily redo transactions which have already

3. output their updates to the database.

 Streamline recovery procedure by periodically performing check pointing

1. Output all log records currently residing in main memory onto stable storage.

2. Output all modified buffer blocks to the disk.

3. Write a log record < checkpoint> onto stable storage.

 During recovery we need to consider only the most recent transaction Ti that started before the
checkpoint, and transactions that started after Ti.
1. Scan backwards from end of log to find the most recent <checkpoint> record

2. Continue scanning backwards till a record <Ti start> is found.

3. Need only consider the part of log following above start record.Earlier part of log can be ignored
during recovery, and can be erased whenever desired.

4. For all transactions (starting from Ti or later) with no <Ti commit>, execute undo(Ti). (Done only
in case of immediate modification.)

5. Scanning forward in the log, for all transactions starting from Ti or later with a <Ti commit>,
execute redo(Ti).

Recovery With Concurrent Transactions

 We modify the log-based recovery schemes to allow multiple transactions to execute concurrently.
 All transactions share a single disk buffer and a single log
 A buffer block can have data items updated by one or more transactions
Dept of CSE ,RGCET Page 27
 DBMS Y3/S5
 We assume concurrency control using strict two-phase locking;
◦ i.e. the updates of uncommitted transactions should not be visible to other
transactions

" Otherwise how to perform undo if T1 updates A, then T2 updates A and

commits, and finally T1 has to abort?

 Logging is done as described earlier.

◦ Log records of different transactions may be interspersed in the log.
 The checkpointing technique and actions taken on recovery have to be changed

 since several transactions may be active when a checkpoint is performed.

 Checkpoints are performed as before, except that the checkpoint log
record is now of the form < checkpoint L> where L is the list of transactions active at the time of the
checkpoint

 We assume no updates are in progress while the checkpoint is carried out (will relax this later)
 When the system recovers from a crash, it first does the following:
1. Initialize undo-list and redo-list to empty

2. Scan the log backwards from the end, stopping when the first

<checkpoint L> record is found.

For each record found during the backward scan:

# if the record is <Ti commit>, add Ti to redo-list

# if the record is <Ti start>, then if Ti is not in redo-list, add Ti to undo-

list

3. For every Ti in L, if Ti is not in redo-list, add Ti to undo-list

 At this point undo-list consists of incomplete transactions which must be undone, and redo-list
consists of finished transactions that must be redone.
 Recovery now continues as follows:
1. Scan log backwards from most recent record, stopping when

<Ti start> records have been encountered for every Ti in undo-list.

During the scan, perform undo for each log record that belongs

to a transaction in undo-list.

Dept of CSE ,RGCET Page 28

 DBMS Y3/S5
2. Locate the most recent <checkpoint L> record.

3. Scan log forwards from the <checkpoint L> record till the end of the log.

During the scan, perform redo for each log record that belongs

to a transaction on redo-list

Example of Recovery

 Go over the steps of the recovery algorithm on the following log:

<T0 start>

<T0, A, 0, 10>

<T0 commit>

<T1 start>

<T1, B, 0, 10>

<T2 start>

/* Scan in Step 4 stops here */

<T2, C, 0, 10>

<T2, C, 10, 20>

<checkpoint {T1, T2}>

<T3 start>

<T3, A, 10, 20>

<T3, D, 0, 10>

<T3 commit>

Recovery Algorithms

Dept of CSE ,RGCET Page 29

 DBMS Y3/S5
 Recovery algorithms are techniques to ensure database consistency and transaction atomicity and
durability despite failures
 Recovery algorithms have two parts
1. Actions taken during normal transaction processing to ensure enough information exists
to recover from failures

2. Actions taken after a failure to recover the database contents to a state that ensures
atomicity, consistency and durability

 The recovery algorithm described in this section requires that a data item that has been updated by
an uncommitted transaction cannot be modified by any other transaction, until the first transaction
has either committed or aborted.
Transaction Rollback

First consider transaction rollback during normal operation (that is, not during recovery from a system
crash).Rollback of a transaction Ti is performed as follows:

1. The log is scanned backward, and for each log record of Ti of the form <Ti , Xj , V1, V2> that is found:

a. The value V1 is written to data item Xj, and

b. Aspecial redo-only log record<Ti , Xj , V1>is written to the log, where V1 is the value being restored to
data item Xj during the rollback. These log records are sometimes called compensation log records.

Such records do not need undo information, since we never need to undo such an undo operation.

2. Once the log record <Ti start> is found the backward scan is stopped, and a log record <Ti abort> is
written to the log.Observe that every update action performed by the transaction or on behalf of the
transaction, including actions taken to restore data items to their old value, have now been recorded in the
log.

Recovery After a System Crash

Recovery actions, when the database system is restarted after a crash, take place in two phases:

1. In the redo phase, the system replays updates of all transactions by scanning the log forward from the
last checkpoint. The log records that are replayed include log records for transactions that were rolled back
before system crash, and those that had not committed when the system crash occurred.

This phase also determines all transactions that were incomplete at the time of the crash, and must
therefore be rolled back. Such incomplete transactions would either have been active at the time of the
checkpoint, and thus would appear in the transaction list in the checkpoint record, or would have started

Dept of CSE ,RGCET Page 30

 DBMS Y3/S5
later; further, such incomplete transactions would have neither a<Ti abort> nor a <Ti commit> record in the
log.

The specific steps taken while scanning the log are as follows:

a. The list of transactions to be rolled back, undo-list, is initially set to the list L in the <checkpoint L> log
record.

b. Whenever a normal log record of the form <Ti , Xj , V1, V2>, or a redo-only log record of the form <Ti , Xj ,
V2> is encountered, the operation is redone; that is, the value V2 is written to data item Xj .

c. Whenever a log record of the form <Ti start> is found, Ti is added to undo-list.

d. Whenever a log record of the form <Ti abort> or <Ti commit> is found, Ti is removed from undo-list.

At the end of the redo phase, undo-list contains the list of all transactions that are incomplete, that is, they
neither committed nor completed rollback before the crash.

2. In the undo phase, the system rolls back all transactions in the undo-list. It performs rollback by scanning
the log backward from the end.

a. Whenever it finds a log record belonging to a transaction in the undolist, it performs undo actions just as
if the log record had been found during the rollback of a failed transaction.

b. When the system finds a <Ti start> log record for a transaction Ti in undo-list, it writes a <Ti abort> log
record to the log, and removes Ti from undo-list.

c. The undo phase terminates once undo-list becomes empty, that is, the system has found <Ti start> log
records for all transactions that were initially in undo-list.

After the undo phase of recovery terminates, normal transaction processing can resume.

Observe that the redo phase replays every log record since the most recent checkpoint record.

The actions include actions of incomplete transactions and the actions carried out to rollback failed
transactions.

The actions are repeated in the same order in which they were originally carried out; hence, this process is
called repeating history.

Figure : shows an example of actions logged during normal operation, and actions performed during failure
recovery. In the log shown in the figure, transaction T1 had committed, and transaction T0 had been
completely rolled back, before the system crashed. Observe how the value of data item B is restored during

Dept of CSE ,RGCET Page 31

 DBMS Y3/S5
the rollback of T0. Observe also the checkpoint record, with the list of active transactions containing T0 and
T1.

When recovering from a crash, in the redo phase, the system performs a redo of all operations after the
last checkpoint record. In this phase, the list undo-list initially contains T0 and T1; T1 is removed first when
its commit log record is found, while T2 is added when its start log record is found. Transaction T0 is undo-
list. The undo phase scans the log backwards from the end, and when it finds a log record of T2 updating A,
the old value of Ais restored, and a redo-only log recordwritten to the log. When the start record for T2 is
found, an abort record is added for T2. Since undo-list contains no more transactions, the undo phase
terminates, completing recovery.

BUFFER MANAGEMENT

Log Record Buffering

 Log record buffering: log records are buffered in main memory, instead of of being output directly
to stable storage.
 Log records are output to stable storage when a block of log records in the buffer is full, or a log
force operation is executed.
 Log force is performed to commit a transaction by forcing all its log records (including the commit
record) to stable storage.
 Several log records can thus be output using a single output operation, reducing the I/O cost.

 The rules below must be followed if log records are buffered:

◦ Log records are output to stable storage in the order in which they are created.
 Transaction Ti enters the commit state only when the log record <Ti commit> has been output
Dept of CSE ,RGCET Page 32
 DBMS Y3/S5
to stable storage.
 Before a block of data in main memory is output to the database, all log records pertaining to
data in that block must have been output to stable storage. " This rule is called the write-ahead
logging or WAL rule – Strictly speaking WAL only requires undo information to be output

Database Buffering

 Database maintains an in-memory buffer of data blocks

 When a new block is needed, if buffer is full an existing block needs to be removed from buffer
 If the block chosen for removal has been updated, it must be output to disk
 As a result of the write-ahead logging rule, if a block with uncommitted updates is output to disk,
log records with undo information for the updates are output to the log on stable storage first.
 No updates should be in progress on a block when it is output to disk. Can be ensured as follows.
 Before writing a data item, transaction acquires exclusive lock on block containing the data item
Lock can be released once the write is completed" Such locks held for short duration are called
latches.
 Before a block is output to disk, the system acquires an exclusive latch on the block "
Ensures no update can be in progress on the block

 Database buffer can be implemented either in an area of real main-memory reserved for the
database, or in virtual memory
 Implementing buffer in reserved main-memory has drawbacks:
 Memory is partitioned before-hand between database buffer and applications, limiting flexibility.
 Needs may change, and although operating system knows best how memory should be divided up
at any time, it cannot change the partitioning of memory.

 Database buffers are generally implemented in virtual memory in spite of some drawbacks:
 When operating system needs to evict a page that has been modified, to make space for another
page, the page is written to swap space on disk.
 When database decides to write buffer page to disk, buffer page may be in swap space, and may
have to be read from swap space on disk and output to the database on disk, resulting in extra I/O!
" Known as dual paging problem.
 Ideally when swapping out a database buffer page, operating system should pass control to
database, which in turn outputs pageto database instead of to swap space (making sure to output
log records first) " Dual paging can thus be avoided, but common operating systems do not support
such functionality.

Fuzzy checkpointing is done as follows:

1. Temporarily stop all updates by transactions

2. Write a <checkpoint L> log record and force log to stable storage

3. Note list M of modified buffer blocks

4. Now permit transactions to proceed with their actions

Dept of CSE ,RGCET Page 33
 DBMS Y3/S5
5. Output to disk all modified buffer blocks in list M

# blocks should not be updated while being output

# Follow WAL: all log records pertaining to a block must be output before the block is output

6. Store a pointer to the checkpoint record in a fixed position last_checkpoint on disk

When recovering using a fuzzy checkpoint, start scan from the checkpoint record pointed to by
last_checkpoint

 Log records before last_checkpoint have their updates reflected in database on disk, and need not
be redone.
 Incomplete checkpoints, where system had crashed while performing checkpoint, are handled
safely

Advanced Recovery Techniques

 Support high-concurrency locking techniques, such as those used for B+-tree concurrency control
 Operations like B+-tree insertions and deletions release locks early.
 They cannot be undone by restoring old values (physical undo), since once a lock is released, other
transactions may have updated the B+-tree.
 Instead, insertions (resp. deletions) are undone by executing a deletion (resp. insertion) operation
(known as logical undo).
 For such operations, undo log records should contain the undo operation to be executed called
logical undo logging, in contrast to physical undo logging.

 Redo information is logged physically (that is, new value for each write) even for such operations
Logical redo is very complicated since database state on disk may not be “operation consistent”

Failure with Loss of Nonvolatile Storage

 So far we assumed no loss of non-volatile storage

 Technique similar to checkpointing used to deal with loss of non-volatile storage
 Periodically dump the entire content of the database to stable storage
 No transaction may be active during the dump procedure; a proceduresimilar to checkpointing
must take place " Output all log records currently residing in main memory onto stable storage.

Dept of CSE ,RGCET Page 34

 DBMS Y3/S5
" Output all buffer blocks onto the disk.

" Copy the contents of the database to stable storage.

" Output a record <dump> to log on stable storage.

 To recover from disk failure" restore database from most recent dump. " Consult the log and redo
all transactions that committed after the dump
 Can be extended to allow transactions to be active during dump; known as fuzzy dump or online
dump
 A dump of the database contents is also referred to as an archival dump, since we can archive the
dumps and use them later to examine old states of the database. Dumps of a database and
checkpointing of buffers are similar.
 Most database systems also support an SQL dump,which writes out SQL DDL statements and SQL
insert statements to a file, which can then be reexecuted to re-create the database.
Fuzzy dump schemes have been developed that allow transactions to be active while the dump is in
progress. They are similar to fuzzy-checkpointing schemes; see the bibliographical notes for more details.

Early Lock Release and Logical Undo Operations

As a result of early lock release, it is possible that a value in a B+-tree node is updated by one transaction
T1, which inserts an entry (V1, R1), and subsequently by another transaction T2, which inserts an entry (V2,
R2) in the same node,moving the entry (V1, R1) even before T1 completes execution.4 At this point, we
cannot undo transaction T1 by replacing the contents of the node with the old value prior to T1 performing
its insert, since that would also undo the insert performed by T2; transaction T2 may still commit (or may
have already committed). In this example, the only way to undo the effect of insertion of (V1, R1) is to
execute a corresponding delete operation.

Logical Operations

The insertion and deletion operations are examples of a class of operations that require logical undo
operations since they release locks early; we call such operations logical operations

Operations acquire lower-level lockswhile they execute, but release them when they complete; the
corresponding transaction must however retain a higher-level lock in a two-phase manner to prevent
concurrent transactions from executing conflicting actions.

Once the lower-level lock is released, the operation cannot be undone by using the old values of updated
data items, and must instead be undone by executing a compensating operation; such an operation is
called a logical undo operation.

Logical Undo Log Records

 Operation logging is done as follows:

Dept of CSE ,RGCET Page 35

 DBMS Y3/S5
1. When operation starts, log <Ti, Oj, operation-begin>. Here Oj is a unique identifier of the
operation instance.

2. While operation is executing, normal log records with physical redo and physical undo
information are logged.

3. When operation completes, <Ti, Oj, operation-end, U> is logged, where U contains information
needed to perform a logical undo information.

 If crash/rollback occurs before operation completes:

 the operation-end log record is not found, and
 the physical undo information is used to undo operation.
 If crash/rollback occurs after the operation completes:
 the operation-end log record is found, and in this case
 logical undo is performed using U; the physical undo information for
 the operation is ignored.
 Redo of operation (after crash) still uses physical redo information.
For example, if the operation inserted an entry in a B+-tree, the undo information U would indicate that a
deletion operation is to be performed, and would identify the B+-tree and what entry to delete from the
tree. Such logging of information about operations is called logical logging. In contrast, logging of old-value
and new-value information is called physical logging, and the corresponding log records are called physical
log records.

To perform logical redo or undo, the database state on disk must be operation consistent, that is, it should
not have partial effects of any operation.

An operation is said to be idempotent if executing it several times in a row gives the same result as
executing it once.

Transaction Rollback With Logical Undo

Dept of CSE ,RGCET Page 36

 DBMS Y3/S5

Rollback of transaction Ti is done as follows:

!Scan the log backwards

 If a log record <Ti, X, V1, V2> is found, perform the undo and log a special redo-only log record <Ti,
X, V1>.
If a <Ti, Oj, operation-end, U> record is found

"

Rollback the operation logically using the undo information U.

– Updates performed during roll back are logged just like during normal operation execution.

– At the end of the operation rollback, instead of logging an operation-end record, generate a
record

<Ti, Oj, operation-abort>.

"

Skip all preceding log records for Ti until the record <Ti, Oj

operation-begin> is found

 Scan the log backwards (cont.):

If a redo-only record is found ignore it

Dept of CSE ,RGCET Page 37

 DBMS Y3/S5
If a <Ti, Oj, operation-abort> record is found:

skip all preceding log records for Ti until the record

<Ti, Oj, operation-begin> is found.

5. Stop the scan when the record <Ti, start> is found

6. Add a <Ti, abort> record to the log

Concurrency Issues in Logical Undo

It is important that the lower-level locks acquired during an operation are sufficient to perform a
subsequent logical undo of the operation; otherwise concurrent operations that execute during normal
processing may cause problems in the undo-phase.

Dept of CSE ,RGCET Page 38