Physical Design and Development

Er. Nawaraj Bhandari
Database Design &
Developmentf
Topic 7 & 8:
Physical Design 2

Types of Constraints
 Entity integrity
 Referential integrity
 Propagation constraints
 Domain constraints
 Table constraints

Activity – Recap - 1
 What is the Entity integrity rule?
 What is referential integrity?

Activity – Recap - 2
 What is the Entity integrity rule?
 The primary key of an entity cannot contain nulls.
 What is referential integrity?
 If a foreign key contains a value then that value
must refer to an existing tuple in the source table

The Art Suppliers data model - where will
there need
to be constraints?
Order OrderItem Item
Supplier
1 1
1
0...*
0...* 0...*
Customer
1
0...*

Source or Parent Table
Create table customers
(CustomerNo integer(5) not null,
first_name varchar(30),
last_name varchar(30),
primary key emp_no);

Referencing or Child Table
Create table Orders
(OrderID integer(5) not null,
CustomerID integer(5) not null,
OrderDate datetime,
primary key OrderID,
foreign key (CustomerID) references Customer
(CustomerID);

Referential Integrity Constraint – Another
Example
Create table workers
(emp_no integer(5) not null,
first_name varchar(30),
last_name varchar(30),
job_title varchar(30),
age integer(3),
dept_no integer(5),
primary key emp_no,
foreign key (dept_no) references Departments (dept_no)o)

Propagation Constraint
 What happens if we delete a Item from our Art Supply database?
 There are lots of OrderItem records that reference it. What happens to
them?

Options for Propagation
 No action
 Cascade
 Set Default
 Set Null

Table with Propagation Constraint
with no action and cascade
Create Table Item
(ItemID integer NOT NULL,
SupplierID integer NOT NULL,
Price float,
Primary Key (ItemID),
Foreign Key (SupplierID) REFERENCES Supplier(SupplierID)
On delete no action
On update cascade);

with Set Default

with no action and cascade
Create Table Item
(ItemID integer NOT NULL,
SupplierID integer NOT NULL,
Price float,
Primary Key (ItemID),
Foreign Key (SupplierID) REFERENCES Supplier(SupplierID)
On delete set Null
On update set Null);

Domain Constraints
 Product Type could be enforced as...
 a check constraint
 separate domain using Create Domain statement
 as a foreign key to another table

Check Constraint
Create Table ProductType
(ProductTypeID integer NOT NULL,
ProductTypeName varchar (30) NOT NULL,
ProductTypeColour varchar (20),
Primary Key (ProductTypeID)
Check (ProductTypeColour in ‘Red’,’Blue’,’Green’));

As a Separate Domain
Create Domain ProductTypeColour As varchar(20)
Default ‘Red’
Check (Value in (‘Red’,’Blue’,’Green’));
 The table’ProductType’ will set the
ProductTypeColour attribute as this domain
ProductTypeColour
ProductTypeColour BoatType,
Primary Key (ProductTypeID));

As a Separate Table
Create Table ProductTypeColour(
(ProductTypeColourCode Varchar(3),
ProductTypeColourDescription Varchar(20)
Primary Key (ProductTypeColourCode));
With the corresponding Foreign Key in Boat
ProductTypeColour varchar (3),
Primary Key (ProductTypeID)
Foreign Key (ProductTypeColourCode)
References ProductTypeColour (ProductTypeColourCode));

Business Rules Enforced by Constraints
 Business rules are derived from requirements analysis and
documented in logical design
 They depend on the operations of the ‘real world’ business

Business Rule: No order may contain
more than 10 OrderItems
Order OrderItem Item
Supplier
1 1
1
0...*
0...10 0...*
Customer
1
0...*

Table Constraints
Create Table Orders
(
(OrderID integer NOT NULL,
CustomerID integer NOT NULL,
OrderDate datetime NOT NULL
Constraint MaximumOrderItems
Check(Not Exists(Select OrderID
From OrderItems
Group By OrderID
Having Count(*) >10)),
Primary Key (OrderID)
Foreign Key (CustomerID) REFERENCES Customer(CustomerID)
On delete no action
On update cascade);

Table Constraints – Another Example
Create Table Rental
(
(BoatID integer NOT NULL,
CustomerID integer NOT NULL,
RentalStartDate datetime NOT NULL,
RentalEndDate datetime NOT NULL
Constraint MaximumRentals
Check(Not Exists(Select BoatID
From Rentals
Group By BoatID
Having Count(*) >10)),
Primary Key (BoatID, CustomerID, RentalStartDate)
Foreign Key (CustomerID) REFERENCES Customer(CustomerID),
Foreign Key (BoatID) REFERENCES Boat (BoatID)
On delete no action
On update cascade);

Some Additional Database Structures
 Views
 Indexes
 Sequences

View of
selected rows
or columns of
these tables
Improving Performance with the use of
Views
Table 1
Table 2
Table 3
Query

Example of Creating a View
Create View OrderSummary as
(Select O.OrderID, O.OrderDate, I.ItemName, I.Price, OI.Quantity, O.Total
From Orders O, OrderItems OI, Items I
Where O.OrderID = OrderItems.OrderID
And I.ItemID = OI.ItemID);
 What will be the purpose of this view?

Indexes
 Improve performance.
 They work by creating entries in a special structure that makes it
easier to find a record

Index Types
 Primary index
 Secondary index
 Clustering index

Primary Index
 Built around a key field that is used for ordering. A unique value for
every entry in the index.

Secondary Index
 Defined on a non-ordering field
 May not contain unique values
 Improves the performance of queries that use columns other than
primary key

Use of Secondary Indexes
 Mechanism for specifying additional key columns.
 For example Customer would be searched often on Customer Name as
well as the primary key and so a secondary index could be used on it.

Examples of Creating Indexes in SQL
 To create primary index
 CREATE UNIQUE INDEX CustomerIDIndex
 ON Customer(CustomerID)
 To create clustering index
 CREATE INDEX OrderDateIndex
 ON Order(OrderDate) CLUSTER( for Oracle)
 To create secondary index use CREATE INDEX syntax without specifying
unique.

Clustered Indexes
 A clustered index alters the way that the rows are physically stored.
 When you create a clustered index on a column the database server
sorts the table’s rows by that column(s).
 It is like a dictionary, where all words are sorted in an alphabetical order.
 (**) Note, that only one clustered index can be created per table i.e.
Primarary Key. It alters the way the table is physically stored, it couldn’t
be otherwise.

Non-Clustered Indexes
 It creates a completely different object within the table, that contains the
column(s) selected for indexing and a pointer back to the table’s rows
containing the data.
 It is like an index in the last pages of a book. All keywords are sorted and
contain a reference back to the appropriate page number. A non-
clustered index on the computer_id column, in the previous example,
would look like the table below:

Clustered Vs Non-Clustered Indexes

Overheads of Use of Indexes
 A record is added to the index table every time a new record is added
to the table where there is a secondary index.
 Updating the indexed record means an update to the index table
 More disk space need to store index tables
 Impact on performance if indexes are all consulted every time a query
is run

Sequences
 Sequential numbers that can be used
to increment an ID number
 Equivalent to an auto-number(Identity)
in MS-Sql Server.

De-normalisation
 We have created a database following all the rules of normalisation...
 Now we can break them to make the database work quicker and
perform better...

Roles in a System
 Not every user is the same
 The will need to access different parts of the system and access it in
different ways

SQL Facilities to Manage Roles
 Grant – gives access to an object (such as a table) to a particular role
or user in the database system.
 Revoke – removes access to an object (such as a table) to a particular
role or user in the database system

Grant
 Grant create on Customers to Admin
 Grant all on Customers to Manager

Revoke
 Revoke all on Customers from Admin
 Revoke delete on Customers to Manager

Past Questions Revision
Why does physical design require an understanding of the chosen
DBMS product?
 In physical design the database is designed with the target
Database Management System (DBMS) in mind. Therefore it
entails knowledge of the chosen DBMS whether it is Oracle, MySQL,
SQL Server etc. The particular features of the chosen system might
specify how structures are set up and stored physically new
features get added and older features become obsolete.

 Define the types of information that should be recorded for each table and column at the physical design stage.
 For each table:
• Table name
• List of columns
• Primary key and foreign keys. Any alternate keys
• Referential integrity constraints for each of the foreign keys that have been identified
 For each column:
• The domain of that column including the data-type, length and any additional
constraints that apply to that column
• A default value for the column
• Whether the column is derived and if it is derived then how it is computed
• Whether or not the column can contain null values

Explain the concept of derived data.:
 Derived data is defined as a column whose value is derived
from the value of one or more other columns in the database.
These might be columns within the same table or columns
from one or more other tables.

How can an organization's derived data make the process of
normalization and data design.
 In an organization there is usually a lot of derived data such
as totals for the number of products ordered.
 When looking at documents during normalization we
sometimes come across derived data so we must make sure
that we map its derivation rather than simply duplicating it..
With regard to design there are similar problems to those in
normalization. Also the derived columns will need to be
modeled. Decisions as to how they are modeled need to take
into account aspects like performance.

What role can SQL views play with regard to derived data?
 Views are ways of storing queries across tables They can be
used to store the results of calculations such as those used
with aggregate functions.

Explain with an example the function of database triggers
 Database triggers are pieces of procedural logic that are
attached to database objects that operate (‘fire’) upon some
event happening such as a new row being inserted into the
database.

 Discuss the ways in which a business rule can be enforced in a
database management system.
 Built into the structure of the database through the normal
constraints e.g. a business rule like ‘Every person must have an
address’ would mean that address fields are specified as not null.
 A rule like ‘A student must be enrolled on a course’ would be a
non-null referential integrity constraint (foreign key).
 More complex business rules can be enforced with check
constraints. Applications can also be used to enforce business
rules through the logic of whatever programming language they are
built in.
 Database triggers can also be used to enforce business rules.

 Explain what an index is in a database system.
 Indexes in databases operate in much the same way as an
index in a book. They are separate files that contain location
pointers to the actual rows in the database.
 Instead of a topic title like in a book index they usually
contain one or more attributes that belong to a row.
 They could, for example, contain just customer numbers from
a customer table, rather than all the data for the customers
as the actual database contains.

 Identify and describe THREE (3) different types of index.
 The different types of indexes available are primary index,
secondary index and clustering index.
 A primary index is built around a key field with a unique value
for every entry.
 A secondary index is used for specifying additional key
columns.
 A clustering index is built around a field which can have many
corresponding rows in the table.

 What is a sequence in a database system and what is it used
for?
 A sequence is a structure that generates a number one after
the other. It is used to populate fields such serial numbers

 Outline the process of enforcing referential integrity in a
database system.
 Foreign keys are identified during the design stage of
development when relationships on an ER diagram are
mapped between foreign keys and their referenced parent key
 It is enforced by the use of constraints known as referential
integrity constraints that are created either when the table is
created or later using he alter table command

 With the use of an example define and explain the purpose of
de-normalisation
 De-normalisation is another way of improving performance
De-normalisation involves using replication in a controlled
way.
 Attributes that do not belong to a particular table but are
often retrieved with that table in queries are replicated on that
that table.

Physical Design and Development

More Related Content

Physical Design and Development

Editor's Notes