Lecture Six
SQL Data Definition
Based on Chapter Six of this book:
Database Systems: A Practical Approach
to Design, Implementation and
Management
International Computer Science S.
Carolyn Begg, Thomas Connolly
Lecture - Objectives
Data types supported by SQL standard.
Purpose of integrity enhancement feature of SQL.
How to define integrity constraints using SQL.
How to use the integrity enhancement feature in
the CREATE and ALTER TABLE statements.
2
Lecture - Objectives
Purpose of views.
How to create and delete views using SQL.
How the DBMS performs operations on views.
Under what conditions views are updatable.
Advantages and disadvantages of views.
How the ISO transaction model works.
3
ISO SQL Data Types
4
Integrity Enhancement Feature
Consider five types of integrity constraints:
Required data.
Domain constraints.
Entity integrity.
Referential integrity.
Enterprise constraints.
5
Integrity Enhancement Feature
Required Data
position VARCHAR(10) NOT NULL
Domain Constraints
(a) CHECK
sex CHAR NOT NULL
CHECK (sex IN ('M', 'F'))
6
Integrity Enhancement Feature
(b) CREATE DOMAIN
CREATE DOMAIN DomainName [AS] dataType
[DEFAULT defaultOption]
[CHECK (searchCondition)]
For example:
CREATE DOMAIN SexType AS CHAR
CHECK (VALUE IN ('M', 'F'));
sex SexType NOT NULL
7
Integrity Enhancement Feature
searchCondition can involve a table lookup:
CREATE DOMAIN BranchNo AS CHAR(4)
CHECK (VALUE IN (SELECT branchNo
FROM Branch));
Domains can be removed using DROP DOMAIN:
DROP DOMAIN DomainName
[RESTRICT | CASCADE]
8
IEF - Entity Integrity
Primary key of a table must contain a unique, non-null
value for each row.
ISO standard supports PRIMARY KEY clause in
CREATE and ALTER TABLE statements:
PRIMARY KEY(staffNo)
PRIMARY KEY(clientNo, propertyNo)
Can only have one PRIMARY KEY clause per table. Can
still ensure uniqueness for alternate keys using UNIQUE:
UNIQUE(telNo)
9
IEF - Referential Integrity
FK is column or set of columns that links each row in
child table containing foreign FK to row of parent table
containing matching PK.
Referential integrity means that, if FK contains a value,
that value must refer to existing row in parent table.
ISO standard supports definition of FKs with FOREIGN
KEY clause in CREATE and ALTER TABLE:
FOREIGN KEY(branchNo) REFERENCES Branch
10
IEF - Referential Integrity
Any INSERT/UPDATE that attempts to create FK value in
child table without matching candidate key value in parent
is rejected.
Action taken that attempts to update/delete a candidate
key value in parent table with matching rows in child is
dependent on referential action specified using ON
UPDATE and ON DELETE subclauses:
CASCADE - SET NULL,
SET DEFAULT - NO ACTION
11
IEF - Enterprise Constraints
Could use CHECK/UNIQUE in CREATE and
ALTER TABLE.
Also have:
CREATE ASSERTION AssertionName
CHECK (searchCondition)
which is very similar to the CHECK clause.
12
IEF - Enterprise Constraints
CREATE ASSERTION StaffNotHandlingTooMuch
CHECK (NOT EXISTS (SELECT staffNo
FROM PropertyForRent
GROUP BY staffNo
HAVING COUNT(*) > 100))
13
Data Definition
SQL DDL allows database objects such as schemas, domains,
tables,views,andindexestobecreatedanddestroyed.
MainSQLDDLstatementsare:
CREATESCHEMA DROPSCHEMA
CREATE/ALTERDOMAIN DROPDOMAIN
CREATE/ALTERTABLE DROPTABLE
CREATEVIEW DROPVIEW
ManyDBMSsalsoprovide:
CREATEINDEX DROPINDEX
14
Data Definition
Relations and other database objects exist in an
environment.
Each environment contains one or more catalogs,
and each catalog consists of set of schemas.
Schema is named collection of related database
objects.
Objects in a schema can be tables, views, domains,
assertions, collations, translations, and character
sets. All have same owner.
15
CREATE SCHEMA
CREATE SCHEMA [Name |
AUTHORIZATION CreatorId ]
DROP SCHEMA Name [RESTRICT | CASCADE ]
With RESTRICT (default), schema must be
empty or operation fails.
With CASCADE, operation cascades to drop
all objects associated with schema in order
defined above. If any of these operations fail,
DROP SCHEMA fails.
16
CREATE TABLE
CREATE TABLE TableName
{(colName dataType [NOT NULL] [UNIQUE]
[DEFAULT defaultOption]
[CHECK searchCondition] [,...]}
[PRIMARY KEY (listOfColumns),]
{[UNIQUE (listOfColumns),] [,]}
{[FOREIGN KEY (listOfFKColumns)
REFERENCES ParentTableName [(listOfCKColumns)],
[ON UPDATE referentialAction]
[ON DELETE referentialAction ]] [,]}
{[CHECK searchCondition)] [,] })
17
CREATE TABLE
Creates a table with one or more columns of the
specified dataType.
With NOT NULL, system rejects any attempt to
insert a row with a null in that column.
Can specify a DEFAULT value for the column.
Primary keys should always be specified as NOT
NULL.
FOREIGN KEY clause specifies FK along with the
referential action
18
Example 6.1 - CREATE TABLE
CREATE DOMAIN OwnerNumber AS VARCHAR(5)
CHECK (VALUE IN (SELECT ownerNo FROM Owner));
CREATE DOMAIN StaffNumber AS VARCHAR(5)
CHECK (VALUE IN (SELECT staffNo FROM Staff));
CREATE DOMAIN PNumber AS VARCHAR(5);
CREATE DOMAIN PRooms AS SMALLINT;
CHECK(VALUE BETWEEN 1 AND 15);
CREATE DOMAIN PRent AS DECIMAL(6,2)
CHECK(VALUE BETWEEN 0 AND 9999.99);
19
Example 6.1 - CREATE TABLE
CREATE TABLE PropertyForRent (
propertyNo PNumber NOT NULL,
.
roomsPRooms NOT NULL DEFAULT 4,
rent PRent NOT NULL, DEFAULT 600,
ownerNo OwnerNumber NOT NULL,
staffNo StaffNumber
Constraint StaffNotHandlingTooMuch
.
branchNo BranchNumber NOT NULL,
PRIMARY KEY (propertyNo),
FOREIGN KEY (staffNo) REFERENCES Staff .);
20
ALTER TABLE
Add a new column to a table.
Drop a column from a table.
Add a new table constraint.
Drop a table constraint.
Set a default for a column.
Drop a default for a column.
21
Example 6.2(a) - ALTER TABLE
Change Staff table by removing default of
'Assistant' for position column and setting default
for sex column to female ('F').
ALTER TABLE Staff
ALTER position DROP DEFAULT;
ALTER TABLE Staff
ALTER sex SET DEFAULT 'F';
22
Example 6.2(b) - ALTER TABLE
Remove constraint from PropertyForRent that
staff not allowed to handle more than 100
properties at time. Add new column to Client table.
ALTER TABLE PropertyForRent
DROP CONSTRAINT StaffNotHandlingTooMuch;
ALTER TABLE Client
ADD prefNoRooms PRooms;
23
DROP TABLE
DROP TABLE TableName [RESTRICT | CASCADE]
e.g. DROP TABLE PropertyForRent;
Removes named table and all rows within it.
With RESTRICT, if any other objects depend for
their existence on continued existence of this table,
SQL does not allow request.
With CASCADE, SQL drops all dependent objects
(and objects dependent on these objects).
24
Views
View
Dynamic result of one or more relational
operations operating on base tables to produce
another table.
Virtual table that does not necessarily actually
exist in the database but is produced upon request,
at time of request.
25
Views
Contents of a view are defined as an SQL
SELECT statement on one or more base tables.
With view resolution, any operations on view are
automatically translated into operations on tables
from which it is derived.
With view materialization, the view is stored as a
temporary table, which is maintained as the
underlying base tables are updated.
26
SQL - CREATE VIEW
CREATE VIEW ViewName [ (newColumnName [,...]) ]
AS subselect
[WITH [CASCADED | LOCAL] CHECK OPTION]
Can assign a name to each column in view.
If list of column names is specified, it must have
same number of items as number of columns
produced by subselect.
If omitted, each column takes name of
corresponding column in subselect.
27
SQL - CREATE VIEW
List must be specified if there is any ambiguity in
a column name.
The subselect is known as the defining query.
WITH CHECK OPTION ensures that if a row
fails to satisfy WHERE clause of defining query, it
is not added to underlying base table.
Need SELECT privilege on all tables referenced in
subselect and USAGE privilege on any domains
used in referenced columns.
28
Example 6.3 - Create Horizontal View
Create view so that manager at branch B003
can only see details for staff who work in his or
her office.
CREATE VIEW Manager3Staff
AS SELECT *
FROM Staff
WHERE branchNo = 'B003';
29
Example 6.4 - Create Vertical View
Create view of staff details at branch B003
excluding salaries.
CREATE VIEW Staff3
AS SELECT staffNo, fName, lName, position, sex
FROM Staff
WHERE branchNo = 'B003';
30
Example 6.5 - Grouped and Joined Views
Create view of staff who manage properties for
rent, including branch number they work at, staff
number, and number of properties they manage.
CREATE VIEW StaffPropCnt (branchNo, staffNo, cnt)
AS SELECT s.branchNo, s.staffNo, COUNT(*)
FROM Staff s, PropertyForRent p
WHERE s.staffNo = p.staffNo
GROUP BY s.branchNo, s.staffNo;
31
Example 6.3 - Grouped and Joined Views
32
SQL - DROP VIEW
DROP VIEW ViewName [RESTRICT | CASCADE]
Causes definition of view to be deleted from
database.
For example:
DROP VIEW Manager3Staff;
33
SQL - DROP VIEW
With CASCADE, all related dependent objects are
deleted; i.e. any views defined on view being
dropped.
With RESTRICT (default), if any other objects
depend for their existence on continued existence
of view being dropped, command is rejected.
34
Restrictions on Views
SQL imposes several restrictions on creation and
use of views.
(a) If column in view is based on an aggregate
function:
Column may appear only in SELECT and
ORDER BY clauses of queries that access view.
Column may not be used in WHERE nor be an
argument to an aggregate function in any query
based on view.
35
Restrictions on Views
For example, following query would fail:
SELECT COUNT(cnt)
FROM StaffPropCnt;
Similarly, following query would also fail:
SELECT *
FROM StaffPropCnt
WHERE cnt > 2;
36
Restrictions on Views
(b) Grouped view may never be joined with a base
table or a view.
For example, StaffPropCnt view is a grouped
view, so any attempt to join this view with
another table or view fails.
37
View Updatability
All updates to base table reflected in all views that
encompass base table.
Similarly, may expect that if view is updated then
base table(s) will reflect change.
38
View Updatability
However, consider again view StaffPropCnt.
If we tried to insert record showing that at branch
B003, SG5 manages 2 properties:
INSERT INTO StaffPropCnt
VALUES ('B003', 'SG5', 2);
Have to insert 2 records into PropertyForRent
showing which properties SG5 manages. However, do
not know which properties they are; i.e. do not know
primary keys!
39
View Updatability
If change definition of view and replace count with
actual property numbers:
CREATE VIEW StaffPropList (branchNo,
staffNo, propertyNo)
AS SELECT s.branchNo, s.staffNo, p.propertyNo
FROM Staff s, PropertyForRent p
WHERE s.staffNo = p.staffNo;
40
View Updatability
Now try to insert the record:
INSERT INTO StaffPropList
VALUES ('B003', 'SG5', 'PG19');
Still problem, because in PropertyForRent all
columns except postcode/staffNo are not allowed
nulls.
However, have no way of giving remaining non-null
columns values.
41
View Updatability
ISO specifies the views that must be updatable in system
that conforms to standard.
A view is updatable iff:
- DISTINCT is not specified.
- Every element in SELECT list of defining query is a column name
and no column appears more than once.
- FROM clause specifies only one table, excluding any views based on
a join, union, intersection or difference.
- No nested SELECT referencing outer table.
- No GROUP BY or HAVING clause.
- Also, every row added through view must not violate integrity
constraints of base table.
42
Updatable View
For view to be updatable, DBMS must be able to
trace any row or column back to its row or
column in the source table.
43
WITH CHECK OPTION
Rows exist in a view because they satisfy WHERE
condition of defining query.
If a row changes and no longer satisfies condition, it
disappears from the view.
New rows appear within view when insert/update on
view cause them to satisfy WHERE condition.
Rows that enter or leave a view are called migrating
rows.
WITH CHECK OPTION prohibits a row migrating
out of the view.
44
WITH CHECK OPTION
LOCAL/CASCADED apply to view hierarchies.
With LOCAL, any row insert/update on view and
any view directly or indirectly defined on this view
must not cause row to disappear from view unless
row also disappears from derived view/table.
With CASCADED (default), any row insert/
update on this view and on any view directly or
indirectly defined on this view must not cause row
to disappear from the view.
45
Example 6.6 - WITH CHECK OPTION
CREATE VIEW Manager3Staff
AS SELECT *
FROM Staff
WHERE branchNo = 'B003'
WITH CHECK OPTION;
Cannot update branch number of row B003 to B002
as this would cause row to migrate from view.
Also cannot insert a row into view with a branch
number that does not equal B003.
46
Example 6.6 - WITH CHECK OPTION
If Manager3Staff is defined not on Staff directly but on
another view of Staff:
CREATE VIEW LowSalary
AS SELECT * FROM Staff WHERE salary > 9000;
CREATE VIEW HighSalary
AS SELECT * FROM LowSalary
WHERE salary > 10000
WITH LOCAL CHECK OPTION;
CREATE VIEW Manager3Staff
AS SELECT * FROM HighSalary
WHERE branchNo = 'B003';
47
Example 6.6 - WITH CHECK OPTION
UPDATE Manager3Staff
SET salary = 9500
WHERE staffNo = 'SG37';
Update would fail: although update would cause
row to disappear from HighSalary, row would not
disappear from LowSalary.
However, if update tried to set salary to 8000,
update would succeed as row would no longer be
part of LowSalary.
48
Example 6.6 - WITH CHECK OPTION
If HighSalary had specified WITH CASCADED
CHECK OPTION, setting salary to 9500 or 8000
would be rejected because row would disappear
from HighSalary.
To prevent anomalies like this, each view should
be created using WITH CASCADED CHECK
OPTION.
49
Advantages of Views
Data Independence
Currency
Improved Security
Reduced Complexity
Convenience
Customization
Data integrity
50
Disadvantages of Views
Update restriction
Structure restriction
Performance
51
Transactions
SQL defines transaction model based on COMMIT and
ROLLBACK.
Transaction is logical unit of work with one or more
SQL statements guaranteed to be atomic with respect
to recovery.
An SQL transaction automatically begins with a
transaction-initiating SQL statement (e.g., SELECT,
INSERT).
Changes made by transaction are not visible to other
concurrently executing transactions until transaction
completes.
52
Transactions
Transaction can complete in one of four ways:
- COMMIT ends transaction successfully, making
changes permanent.
- ROLLBACK aborts transaction, backing out any
changes made by transaction.
- For programmatic SQL, successful program
termination ends final transaction successfully, even
if COMMIT has not been executed.
- For programmatic SQL, abnormal program end
aborts transaction.
53
Transactions
New transaction starts with next transaction-
initiating statement.
SQL transactions cannot be nested.
54
