Download Relation

The Relational Model
Lecture Topics
• Overview of SQL
• Underlying relational model
• Relational database structure
• SQL DDL and DML
• Views
Textbook
• Chapter 3
CS 338
The Relational Model
2-1
Overview of SQL
• Structured Query Language (SQL,
sometimes pronounced “sequel”)
• ISO/IEC 9075, international standard
for relational database systems
• the standard is evolving:
– 1986: SQL1; initial version, 75 pages
– 1989: SQL89; most commercial
products conform to this version, with
extensions
– 1992: SQL2; three levels of
conformity, ~600 pages
– 1999: SQL99 (sometimes called
SQL3), ~1200 pages
– 2007: under revision again
CS 338
The Relational Model
2-2
...continued
Main features:
• powerful view definition language
• integrity constraints in conceptual
schema
• DML can be embedded in various
programming languages (called
“embedded SQL”), or used via
programming libraries
– object/class libraries in OO
environments
• transaction control
• authorization sublanguage/model
CS 338
The Relational Model
2-3
...continued
Application 1
D
B
M
S
S
Q
L
Application 2
View A
Application 3
View B
Conceptual Level
Physical Schema
Database
CS 338
The Relational Model
2-4
Underlying relational
model
Example relational database for a credit card company
Vendor
Vno Vname
1 Sears
2 Kmart
3 Esso
4 Esso
City
Toronto
Ottawa
Montreal
Waterloo
Customer
AccNum Cname
101
Smith
102
Jones
103
Martin
Transaction
Tno
Vno
1001 2
1002 2
1003 3
1004 4
1005 4
CS 338
Prov
Ont
BC
Que
AccNum
101
103
101
102
103
Vbal
200.00
671.05
0.00
2.25
Cbal
25.15
2014.00
150.00
Tdate
070115
070116
070115
070120
070125
The Relational Model
Climit
2000
2500
1000
Amount
13.25
19.00
25.00
16.13
33.12
2-5
Structure of a relational
database
Database: collection of uniquely named
tables (relations)
Relation: set of rows (tuples)
Attribute: column
Domain: set of allowed values for an
attribute
Attribute values must be atomic (single
values): no tuples or sets or repetition
Row: distinguishable thing
Table: set of related things
CS 338
The Relational Model
2-6
Diagrammatic conventions
Vendor
Vno Vname City
Vbal
or
Vendor
Vno
Vname
City
Vbal
CS 338
The Relational Model
2-7
Pictorial schema
Customer
AccNum
Cname
Prov
Cbal
Climit
Vendor
Vno
Vname
City
VBal
CS 338
Transaction
Tno
Vno
AccNum
Tdate
Amount
The Relational Model
2-8
The SQL DDL
• used for defining
– tables
– views
• example of table definition (conceptual
schema):
create table Vendor
(Vno
INTEGER not null,
Vname
VARCHAR(20),
City
VARCHAR(10),
Vbal
DECIMAL(10,2),
primary key (Vno) );
CS 338
The Relational Model
2-9
...continued
create table Customer
(AccNum INTEGER not null,
Cname VARCHAR(20) not null,
Prov
VARCHAR(20),
Cbal
DECIMAL(6,2) not null,
Climit DECIMAL(4,0) not null,
primary key (AccNum) );
create table Transaction
(Tno
INTEGER not null,
Vno
INTEGER not null,
AccNum INTEGER not null,
Tdate DATE,
Amount DECIMAL(6,2) not null,
primary key (Tno),
foreign key (Vno)
references vendor(Vno),
foreign key (AccNum)
references Customer(AccNum) );
CS 338
The Relational Model
2-10
Attribute domains in SQL
• INTEGER: integers representable with
32 bits
• SMALLINT: integers representable
with 16 bits
• DECIMAL(m,n): fixed point numbers
• FLOAT: 32 bit floating point numbers
• CHAR(n): fixed length strings
• VARCHAR(n): variable length strings
• BIT(n): n bits
• BIT VARYING(n): variable number of
bits
CS 338
The Relational Model
2-11
...continued
• DATE (year, month, day)
• TIME (hour, minute, second)
• TIME(i) (hour, minute, second, second
fraction)
• TIMESTAMP (date, time, second
fraction)
• INTERVAL YEAR/MONTH (year month
interval)
• INTERVAL DAY/TIME (day time interval)
• plus many, many product-specific (nonstandard) extensions
CS 338
The Relational Model
2-12
Modifying table definitions
• Table schemas can be changed after
the table has been created:
–
–
–
–
adding columns
removing columns
removing constraints (e.g. p-key)
some SQL implementations allow
• renaming a column
• modifying a column
• Example:
ALTER TABLE Vendor
ADD Street VARCHAR(15)
CS 338
The Relational Model
2-13
Removing tables
• SQL operation is “drop”
• Tables can be dropped at any time
• Dropping a table deletes the schema
and the instance
• All views, foreign-key definitions are
also removed
• Example:
DROP TABLE Transaction
CS 338
The Relational Model
2-14
The SQL DML
SQL has a non-navigational DML:
E.g. “Find names and provinces of
customers who owe more than $1000 to
the company.”
select Cname, Prov
from Customer
where Cbal > 1000;
CS 338
The Relational Model
2-15
...continued
• basic querying:
select columns
from R1,..., Rk
[ where filter ];
• result is a relation over columns
(columns = “*” means all attributes in
R1,..., Rk)
• R1,..., Rk: tables from which the data is
retrieved
• filter : conditions on tuples used to
form the result; optional
CS 338
The Relational Model
2-16
...continued
• conditions may include:
– arithmetic operators +, -, *, /
– comparisons =, <>, <, <=, >, >=
– logical connectives and, or and not
E.g. “List the names of the customers who
live in Ontario and whose balance is over
80% of their balance limit.”
select Cname
from Customer
where Prov = 'Ont' and
Cbal > 0.8 * Climit;
CS 338
The Relational Model
2-17
...continued
• basic insertion:
insert into Customer
values (104, ‘Anne', 'ON',
0, 4000);
• deletion: delete Customer rows for
customers named Smith:
delete from Customer
where Cname = 'Smith';
• delete all transactions:
delete from Transaction;
CS 338
The Relational Model
2-18
...continued
• modification, changing existing rows
• set balance of account 102 to zero:
update Customer set Cbal = 0
where AccNum = 102;
• add $100 to each customer’s monthly
limit:
update Customer set Climit =
Climit + 100;
CS 338
The Relational Model
2-19
SQL external schema
• called views
• a view is a named query (result is usually
computed when the view is used)
create view WatVendors as
select VNo, VName, VBal
from Vendor
where City = 'Waterloo';
• views can be used in retrieval exactly like
tables (but updates of views are
restricted)
CS 338
The Relational Model
2-20
Advantages of views
• logical data independence
• simplified perception of the database
• different views for different users
• restricting data access
CS 338
The Relational Model
2-21
Basic relational concepts
Relating to descriptions of data:
• Attribute (column): a name denoting
a property or characteristic
• Relation schema (table header): a
finite set of attributes and a mapping of
each attribute to a domain (defined
below)
CS 338
The Relational Model
2-22
...continued
Relating to data:
Domain: an “abstract data type” (i.e. a
name, a set of values and a number of
functions defined over the values)
• Null value: a special exception value
(meaning “not known”, “not
applicable”)
• Tuple: a set of attribute/value pairs,
with each attribute occurring at most
once
• Relation (table): a relation schema,
and a finite set of tuples
• Relational database: a finite set of
relation names and a mapping of each
relation name to a relation
CS 338
The Relational Model
2-23
...continued
Other:
• Intention of a relation: the associated
relation schema
• Extension of a relation: the associated
set of tuples
The relational model assumes no ordering
of either rows or columns for any table.
CS 338
The Relational Model
2-24
Basic rules
• Domain constraints: the value
associated with each attribute in a
tuple must occur in the set of values
associated with the domain of the
attribute; or the value is Null
• First normal form: domain values
cannot be tuples or relations
• Completeness: each tuple in a
relation has an attribute/value pair for
precisely the set of attributes in the
associated relation schema
• Closed world: the database “knows
of” all tuples in all relations
• Unique rows: no two distinct tuples in
any given relation consist of the same
set of attribute/value pairs
CS 338
The Relational Model
2-25
Keys
• Relation superkey: a subset of the
associated relation schema for which no
pair of distinct tuples in the relation will
ever agree on the corresponding
values.
• Relation candidate key: a minimal
superkey
• Relation primary key: a distinguished
candidate key of the relation
• Foreign key: primary key of one
relation appearing as attributes of
another relation
• Foreign keys enable capturing more
complex entity structure
CS 338
The Relational Model
2-26
Integrity of primary and
foreign keys
• Entity integrity: No component of a
primary key value may be the null
value, nor may be updated.
• Referential integrity: A tuple with a
non-null value for a foreign key that
does not match the primary key value
of a tuple in the referenced relation is
not allowed.
CS 338
The Relational Model
2-27