Constraints

INTEGRITY CONSTRAINTS
Integrity Constraints are used to prevent entry of invalid information into tables. There
are five Integrity Constraints Available in Oracle. They are :

 Not Null
 Primary Key
 Foreign Key
 Check
 Unique

Not Null
By default all columns in a table can contain null values. If you want to ensure that a
column must always have a value, i.e. it should not be left blank, then define a NOT
NULL constraint on it.

Always be careful in defining NOT NULL constraint on columns, for example in

employee table some employees might have commission and some employees might
not have any commission. If you put NOT NULL constraint on COMM column then you
will not be able insert rows for those employees whose commission is null. Only put NOT
NULL constraint on those column which are essential for example in EMP table ENAME
column is a good candidate for NOT NULL constraint.

Related: NVL() function in Oracle

Primary Key
Each table can have one primary key, which uniquely identifies each row in a table and
ensures that no duplicate rows exist. Use the following guidelines when selecting a
primary key:

 Whenever practical, use a column containing a sequence number. It is a simple

way to satisfy all the other guidelines.
 Minimize your use of composite primary keys. Although composite primary keys
are allowed, they do not satisfy all of the other recommendations. For example,
composite primary key values are long and cannot be assigned by sequence
numbers.
 Choose a column whose data values are unique, because the purpose of a
primary key is to uniquely identify each row of the table.
 Choose a column whose data values are never changed. A primary key value is
only used to identify a row in the table, and its data should never be used for any
other purpose. Therefore, primary key values should rarely or never be changed.
 Choose a column that does not contain any nulls. A PRIMARY KEY constraint,
by definition, does not allow any row to contain a null in any column that is part of
the primary key.
  Choose a column that is short and numeric. Short primary keys are easy to type.
You can use sequence numbers to easily generate numeric primary keys.

For example in EMP table EMPNO column is a good candidate for PRIMARY KEY.

To define a primary key on a table give the following command.

alter table emp add constraint emppk primary key (empno);

The above command will succeed only if the existing values are compliant i.e. no
duplicates are there in EMPNO column. If EMPNO column contains any duplicate value
then the above command fails and Oracle returns an error indicating of non compliant
values.

Whenever you define a PRIMARY KEY Oracle automatically creates a index on that
column. If an Index already exist on that column then Oracle uses that index.

FOREIGN KEY
On whichever column you put FOREIGN KEY constraint then the values in that column
must refer to existing values in the other table. A foreign key column can refer to primary
key or unique key column of other tables. This Primary key and Foreign key relationship
is also known as PARENT-CHILD relationship i.e. the table which has Primary Key is
known as PARENT table and the table which has Foreign key is known as CHILD table.
This relationship is also known as REFERENTIAL INTEGRITY.

The following shows an example of parent child relationship.

Here EMPNO in attendance table is a foreign key referring to EMPNO of EMP table.

alter table attendance add constraint empno_fk

foreign key (empno) references emp(empno);
The above command succeeds only if EMPNO column in ATTENDANCE table contains
values which are existing in EMPNO column of EMP table. If any value is not existing
then the above statement fails and Oracle returns an error indicating non compliant
values.

Some points to remember for referential integrity

 You cannot delete a parent record if any existing child record is there. If you have
to first delete the child record before deleting the parent record. In the above
example you cannot delete row of employee no. 101 since it’s child exist in the
ATTENDANCE table. However, you can delete the row of employee no. 103
since no child record exist for this employee in ATTENDANCE table. If you
define the FOREIGN KEY with ON DELETE CASCADE option then you can
delete the parent record and if any child record exist it will be automatically
deleted.

To define a foreign key constraint with ON DELETE CASCADE option give the following
command.

ALTER TABLE attendance ADD CONSTRAINT empno_fk

FOREIGN KEY (empno) REFERENCES emp(empno)
ON DELETE CASCADE;

From Oracle version 9i, Oracle has also given a new feature i.e. ON DELETE SET NULL
. That is it sets the value for foreign key to null whenever the parent record is deleted.

To define a foreign key constraint with ON DELETE SET NULL option give the following
command.

ALTER TABLE attendance ADD CONSTRAINT empno_fk

FOREIGN KEY (empno) REFERENCES emp(empno)
ON DELETE SET NULL;

 You also cannot drop the parent table without first dropping the FOREIGN KEY
constraint from attendance table. However if you give CASCADE
CONSTRAINTS option in DROP TABLE statement then Oracle will automatically
drop the references and then drops the table.

CHECK
Use the check constraint to validate values entered into a column. For example in the
above ATTENDANCE table, the DAYS column should not contain any value more than
31. For this you can define a CHECK constraint as given below

alter table attendance add constraint dayscheck

check (days <= 31);
Similarly if you want the salaries entered in to SAL column of employee table should be
between 1000 and 20000 then you can define a CHECK constraint on EMP table as
follows

alter table emp add constraint sal_check

check (sal between 1000 and 20000);

You can define as many check constraints on a single column as you want there is no
restrictions on number of check constraints.

UNIQUE KEY
Unique Key constraint is same as primary key i.e. it does not accept duplicate values,
except the following differences

 There can be only on Primary key per table. Whereas, you can have as many
Unique Keys per table as you want.
 Primary key does not accept NULL values whereas, unique key columns can be
left blank.
 You can also refer to Unique key from Foreign key of other tables.

On which columns you should put Unique Key Constraint ?

It depends on situations, first situation is suppose you have already defined a Primary
key constraint on one column and now you have another column which also should not
contain any duplicate values, Since a table can have only one primary key, you can
define Unique Key constraint on these columns. Second situation is when a column
should not contain any duplicate value but it should also be left blank. For example in the
EMP table IDNO is a good candidate for Unique Key because all the IDNO’s are unique
but some employees might not have ID Card so you want to leave this column blank.

To define a UNIQUE KEY constraint on an existing table give the following command.

alter table emp add constraint id_unique unique (idno);

DEFAULT
You can also specify the DEFAULT value for columns i.e. when user does not enter
anything in that column then that column will have the default value. For example in
EMP table suppose most of the employees are from Hyderabad, then you can put this
as default value for CITY column. Then while inserting records if user doesn’t enter
anything in the CITY column then the city column will have Hyderabad.
To define default value for columns create the table as given below

create table emp (empno number(5),

name varchar2(20),
sal number(10,2),
city varchar2(20) default ‘Hyd’);

Now, when user inserts record like this

insert into emp values (101,’Sami’,2000,’Bom’);

Then the city column will have value ‘Bom ‘. But when user inserts a record like this

insert into emp (empno,name,sal) values (102,’Ashi’,4000);

Then the city column will have value ‘Hyd’. Since it is the default.

Examples
Defining Constraints in CREATE TABLE statement.

create table emp (empno number(5) constraint emppk

Primary key,
ename varchar2(20) constraint namenn
not null,
sal number(10,2) constraint salcheck
check (sal between 1000 and 20000)
idno varchar2(20) constraint id_unique
unique );

create table attendance (empno number(5) constraint empfk

references emp (empno)
on delete cascade,
month varchar2(10),
days number(2) constraint dayscheck
check (days <= 31) );

The name of the constraints are optional. If you don’t define the names then oracle
generates the names randomly like ‘SYS_C1234’

Another way of defining constraint in CREATE TABLE statement.

create table emp (empno number(5),

ename varchar2(20) not null,
sal number(10,2),
idno varchar2(20),
constraint emppk Primary key (empno)
constraint salcheck check (sal between 1000 and
20000)
constraint id_unique unique (idno) );
create table attendance (empno number(5),
month varchar2(10),
days number(2),
constraint empfk foreign key (empno)
references emp (empno)
on delete cascade
constraint dayscheck
check (days <= 31) );

Deferring Constraint Checks

You may wish to defer constraint checks on UNIQUE and FOREIGN keys if the data you
are working with has any of the following characteristics:

 Tables are snapshots

 Tables that contain a large amount of data being manipulated by another
application, which may or may not return the data in the same order
 Update cascade operations on foreign keys

When dealing with bulk data being manipulated by outside applications, you can defer
checking constraints for validity until the end of a transaction.

Ensure Constraints Are Created Deferrable

After you have identified and selected the appropriate tables, make sure their FOREIGN,
UNIQUE and PRIMARY key constraints are created deferrable. You can do so by
issuing a statement similar to the following:

create table attendance (empno number(5),

month varchar2(10),
days number(2),
constraint empfk foreign key (empno)
references emp (empno)
on delete cascade
DEFERRABLE
constraint dayscheck
check (days <= 31) );

Now give the following statement

set constraint empfk deferred;

update attendance set empno=104 where empno=102;
insert into emp values (104,’Sami’,4000,’A123’);
commit;

You can check for constraint violations before committing by issuing the SET
CONSTRAINTS ALL IMMEDIATE statement just before issuing the COMMIT. If there
are any problems with a constraint, this statement will fail and the constraint causing the
error will be identified. If you commit while constraints are violated, the transaction will be
rolled back and you will receive an error message.

ENABLING AND DISABLING CONSTRIANTS

You can enable and disable constraints at any time.

To enable and disable constraints the syntax is

ALTER TABLE <TABLE_NAME> ENABLE/DISABLE

CONSTRAINT <CONSTRAINT_NAME>

For example to disable primary key of EMP table give the following statement

alter table emp disable constraint emppk;

And to enable it again, give the following statement

alter table emp enable constraint emppk;

Dropping constraints
You can drop constraint by using ALTER TABLE DROP constraint statement.

For example to drop Unique constraint from emp table, give the following statement

alter table emp drop constraint id_unique;

To drop primary key constraint from emp table.

alter table emp drop constraint emppk;

The above statement will succeed only if the foreign key is first dropped otherwise you
have to first drop the foreign key and then drop the primary key. If you want to drop
primary key along with the foreign key in one statement then CASCADE CONSTRAINT
statement like this

alter table emp drop constraint emppk cascade;

Viewing Information about constraints

To see information about constraints, you can query the following data dictionary tables.
select * from user_constraints;
select * from user_cons_columns;