Languages of DBMS

Data Definition Language DDL

define the schema and storage stored in a Data
Dictionary
Relational Query Languages

Data Manipulation Language DML

Manipulative populate schema, update database

Retrieval querying content of a database

Data Control Language DCL

permissions, access control etc...

Data Manipulation Language
Relational operators transform either a single

relation or a pair of relations into a result that is
a relation that can be used as an operand on

Theory behind operations is formally defined
later operations
and equivalent to a first-order logic (FOL)

For every operator operand and result,

Relational Calculus (∀, ∃ ) ≡ Relational relations are free of duplicates
Algebra
Operators are tuple oriented or set oriented

Relational algebra is a retrieval query language
Structured Query Language (SQL)
based on set operators and relational
an ANSI standard for relational databases,
operators based on relational algebra/calculus

SQL2 1992

SQL3 1998

Operations in the Relational Model

Query Operators

Theory behind operations is formally defined
and equivalent to a first order logic (FOL)
Relational Algebra

Relational operators transform either a simple
tuple (unary) Selection, Projection
relation or a pair of relations into a result that
is a relation
set (binary) Union, Intersection, Difference

tuple (binary) Join, Division

The result can be used as an operand on
Additional Operators
later activities
Outer Join, Outer Union

For every operand and result, relations are
free of duplicates

Operators are tuple oriented or set oriented

1
 Relational Algebra
A Retrieval DML Must Express Select Project

Attributes required in a result

Union

target list Intersection Difference

R
Criteria for selecting tuples for that result R R

S S S

qualifier

The relations that take part in the query Join Division

set generators *

Independent of the instances in the database

Cartesian product
Expressions are in terms of the database
X
schema

SQL Retrieval Statement π Project Operator

SELECT[all|distinct]
selects a subset of the attributes of a relation
{*|{table.*|expr[alias]|view.*}
[,{table.*|expr[alias]}]...} attribute list are drawn from the specified relation;
FROM table [alias][,table[alias]] ... if the key attribute is in the list then card(result) = card(relation)
[WHERE condition]
[CONNECT BY condition
[START WITH condition]] Result = π (attribute list)(relation name)
[GROUP BY expr [,expr] ...]
[HAVING condition] resulting relation
[{UNION|UNION ALL|INTERSECT|MINUS} has only the the degree(result) = number of attributes
attributes in the in the attribute list
SELECT ...]
list, in same order
[ORDER BY {expr|position} as they appear in no duplicates in the result
[ASC|DESC][,expr|position}[ASC|DESC]. the list
[FOR UPDATE OF column [,column] ... [NOWAIT]]

π Project Operator π Project Operator SELECT

STUDENT
STUDENT
studno name hons tutor year
studno name hons tutor year select *
s1 jones ca bush 2
s1 jones ca bush 2 s2 brown cis kahn 2 from student;
s2 brown cis kahn 2 s3 smith cs goble 2
s3 smith cs goble 2 s4 bloggs ca goble 1
s4 bloggs ca goble 1 s5 jones cs zobel 1
s5 jones cs zobel 1 s6 peters ca kahn 3
s6 peters ca kahn 3
tu to r
bush
tutor
kahn select tutor
πtutor(STUDENT)
bush g o b le from student;
kahn g o b le
goble zobel
zobel kahn

2
 σ Select Operator σ Select Operator
selects a subset of the tuples in a relation that
STUDENT
satisfy a selection condition studno name hons tutor year
s1 jones ca bush 2
s2 brown cis kahn 2
a boolean expression specified on the attributes
s3 smith cs goble 2
of a specified relation s4 bloggs ca goble 1
s5 jones cs zobel 1

Result = σ (selection condition)(relation name)

s6 peters ca kahn 3

a relation that • stands for the usual comparison operators ‘<‘, studno name hons tutor year
has the same ‘<>‘, ‘<=‘, ‘>‘, ‘>=‘, etc s4 bloggs ca goble 1
attributes as the • clauses can be arbitrarily connected with
source relation; boolean operators AND, NOT, OR

degree(result) = degree(relation);
σname=‘bloggs’(STUDENT)
card(result) <= card(relation)

retrieve tutor who tutors Bloggs SQL retrieval expressions

STUDENT
studno name hons tutor year
s1 jones ca bush 2
select studentno, name from student
s2 brown cis kahn 2
s3 smith cs goble 2 where hons != ‘ca’ and
s4 bloggs ca goble 1
s5 jones cs zobel 1 (tutor = ‘goble’ or tutor = ‘kahn’);
s6 peters ca kahn 3

select * from enrol
where labmark > 50;
πtutor(σname=‘bloggs’(STUDENT))

select * from enrol
select tutor from student where labmark between 30 and 50;
where name = ‘bloggs’;

Cartesian Product Operator

select * from enrol
where labmark in (0, 100);
Definition:

select * from enrol
The cartesian product of two relations
where labmark is null; R1(A1,A2,...,An) with cardinality i and

select * from student R2(B1,B2,...,Bm) with cardinality j is a relation
R3 with degree k=n+m, cardinality i*j and
where name is like ‘b%’; attributes (A1,A2,...,An,B1,B2,...,Bm)

select studno, courseno,
The result, denoted by R1XR2, is a relation
exammark+labmark total from enrol that includes all the possible combinations of
tuples from R1 and R2
where labmark is not NULL;
Used in conjunction with other operations

3
 Cartesian Product Example X Cartesian Product
STUDENT
STUDENT COURSE
studno name hons tutor year lecturer roomno
studno name hons tutor year s1 jones ca bush 2 kahn IT206
studno name courseno subject equip s1 jones ca bush 2 s1
s1
jones
jones
ca
ca
bush
bush
2
2
bush
goble
2.26
2.82
s1 jones cs250 prog sun s2 brown cis kahn 2 s1 jones ca bush 2 zobel 2.34
s1 jones ca bush 2 watson IT212
s2 brown cs150 prog sun s3 smith cs goble 2 s1 jones ca bush 2 woods IT204
s6 peters cs390 specs sun s4 bloggs ca goble 1 s1 jones ca bush 2 capon A14
s1 jones ca bush 2 lindsey 2.10
s5 jones cs zobel 1 s1 jones ca bush 2 barringer 2.125
s2 brown cis kahn 2 kahn IT206
s6 peters ca kahn 3
STUDENT X COURSE s2
s2
brown
brown
cis
cis
kahn
kahn
2
2
bush
goble
2.26
2.82
studno name courseno subject equip s2 brown cis kahn 2 zobel 2.34
STAFF s2 brown cis kahn 2 watson IT212
s1 jones cs250 prog sun s2 brown cis kahn 2 woods IT204
s1 jones cs150 prog sun lecturer roomno s2 brown cis kahn 2 capon A14
s2 brown cis kahn 2 lindsey 2.10
s1 jones cs390 specs sun kahn IT206 s2 brown cis kahn 2 barringer 2.125
s2 brown cs250 prog sun bush 2.26 s3 smith cs goble 2 kahn IT206
s3 smith cs goble 2 bush 2.26
s2 brown cs150 prog sun goble 2.82 s3 smith cs goble 2 goble 2.82
s2 brown cs390 specs sun zobel 2.34 s3
s3
smith
smith
cs
cs
goble
goble
2
2
zobel
watson
2.34
IT212
s6 peters cs250 prog sun watson IT212 s3 smith cs goble 2 woods IT204
s6 peters cs150 prog sun s3 smith cs goble 2 capon A14
woods IT204 s3 smith cs goble 2 lindsey 2.10
s6 peters cs390 specs sun capon A14 s3 smith cs goble 2 barringer 2.125
s4 bloggs ca goble 1 kahn IT206
lindsey 2.10 s4 bloggs ca goble 1 bush 2.26
barringer 2.125

θ Join Operator θ Join Operator STAFF

lecturer roomno
Definition: STUDENT
studno name hons tutor year
kahn IT206
bush 2.26
The join of two relations R1(A1,A2,...,An) and s1 jones ca bush 2 goble 2.82
R2(B1,B2,...,Bm) is a relation R3 with degree k=n+m s2 brown cis kahn 2 zobel 2.34
s3 smith cs goble 2 watson IT212
and attributes (A1,A2,...,An, B1,B2,...,Bm) that satisfy s4 bloggs ca goble 1 woods IT204
the join condition s5 jones cs zobel 1 capon A14
s6 peters ca kahn 3 lindsey 2.10
Result = R1 (θ join condition) R2 barringer 2.125

stud name hons tutor year lecturer roomno

•stands for the usual comparison no
The result is a concatenated operators ‘<‘, ‘<>‘, ‘<=‘, ‘>‘, ‘>=‘, etc s1 jones ca bush 2 bush 2.26
set but only for those tuples s2 brown cis kahn 2 kahn IT206
where the condition is true. •comparing terms in the Θ clauses can
s3 smith cs goble 2 goble 2.82
It does not require union s4 bloggs ca goble 1 goble 2.82
be arbitrarily connected with boolean
compatibility of R1 and R2 s5 jones cs zobel 1 zobel 2.34
operators AND, NOT, OR s6 peters ca kahn 3 kahn IT206

ENROL
More joins stud course lab exam
no no mark mark
STUDENT s1 cs250 65 52 Natural Join Operator
studno name hons tutor year s1 cs260 80 75
Of all the types of θ-join, the equi-join is the
s1 jones ca bush 2 s1 cs270 47 34
s2 brown cis kahn 2 s2 cs250 67 55

s3 smith cs goble 2 s2 cs270 65 71 only one that yields a result in which the
s4 bloggs ca goble 1 s3 cs270 49 50
s5 jones cs zobel 1 s4 cs280 50 51 compared columns are redundant to each
s6 peters ca kahn 3 s5 cs250 0 3 other—possibly different names but same
s6 cs250 2 7
values

The natural join is an equi-join but one of the
stud name hons tutor year stud course lab exam
no no no mark mark redundant columns (simple or composite) is
s1 jones ca bush 2 s1 cs250 65 52
s1 jones ca bush 2 s1 cs260 80 75 omitted from the result
s1 jones ca bush 2 s1 cs270 47 34
s2 brown cis kahn 2 s2 cs250 67 55
Relational join is the principle algebraic
s2 brown cis kahn 2 s2 cs270 65 71
s3 smith cs goble 2 s3 cs270 49 50 counterpart of queries that involve the
s4
s5
bloggs
jones
ca
cs
goble
zobel
1
1
s4
s5
cs280
cs250
50
0
51
3 existential quantifier ∃
s6 peters ca kahn 3 s6 cs250 2 7

4
 Self Join: Joins on the same relation Exercise
π (lecturer, (staff (appraiser = lecturer) staff) Get student’s name, all their courses, subject
roomno,appraiser, approom) of course, labmark for course, lecturer of
course and lecturer’s roomno for ‘ca’ students
STAFF
lecturer roomno appraiser
lecturer roomno appraiser approom
kahn IT206 watson
bush 2.26 capon
kahn IT206 watson IT212 University Schema
bush 2.26 capon A14
goble 2.82 capon goble 2.82 capon A14
STUDENT(studno,name,hons,tutor,year)
zobel 2.34 watson zobel 2.34 watson IT212
watson IT212 barringer watson IT212 barringer 2.125
ENROL(studno,courseno,labmark,exammark)
woods IT204 barringer woods IT204 barringer 2.125
capon A14 watson capon A14 watson IT212
COURSE(courseno,subject,equip)
lindsey 2.10 woods lindsey 2.10 woods IT204
barringer 2.125 null
STAFF(lecturer,roomno,appraiser)

TEACH(courseno,lecturer)
select e.lecturer, e.roomno, m.lecturer appraiser, m.roomno approom
from staff e, staff m
YEAR(yearno,yeartutor)
where e.appraiser = m.lecturer

Set Theoretic Operators ∪ Union Operator

Definition:

Union, Intersection and Difference The union of two relations

Operands need to be union compatible for R1(A1,A2,...,An) and R2(B1,B2,...,Bm)
the result to be a valid relation is a relation R3(C1,C2,...,Cn) such that
dom(Ci)= dom(Ai) =dom (Bi) for 1 ≤ i ≤ n
Definition:
Two relations
The result R1 ∪ R2 is a relation that includes
R1(A1,A2,...,An) and R2(B1,B2,...,Bm) all tuples that are either in R1 or R2 or in both
are union compatible iff: without duplicate tuples
n = m and, dom(Ai)= dom (Bi) for 1 ≤ i ≤ n
The resulting relation might have the same
attribute names as the first or the second
relation

Retrieve all staff that lecture or tutor ∩ Intersection Operator

STUDENT TEACH
studno name hons tutor year courseno lecturer
s1 jones ca bush 2 cs250
cs250
lindsey
capon Definition:
s2 brown cis kahn 2
s3 smith cs goble 2 cs260 kahn The intersection of two relations
cs260 bush
s4 bloggs ca goble 1
cs270 zobel
R1(A1,A2,...,An) and R2(B1,B2,...,Bm) is a
s5 jones cs zobel 1
s6 peters ca kahn 3 cs270 woods relation R3(C1,C2,...,Cn) such that
dom(Ci)= dom(Ai) ∩ dom (Bi) for 1 ≤ i ≤ n
cs280 capon

lecturer
lindsey
Lecturers π(lecturer)TEACH capon
The result R1 ∩ R2 is a relation that includes
kahn only those tuples in R1 that also appear in R2
Tutors π(tutor)STUDENT bush
zobel
The resulting relation might have the same
Lecturers ∪ Tutors woods
capon attribute names as the first or the second
goble relation

5
Retrieve all staff that lecture and tutor − Difference Operator
STUDENT TEACH
studno name
s1 jones
hons
ca
tutor
bush
year
2
courseno
cs250
lecturer
lindsey
Definition:
s2 brown cis kahn 2 cs250 capon The difference of two relations
cs260 kahn
s3
s4
smith
bloggs
cs
ca
goble
goble
2
1 cs260 bush
R1(A1,A2,...,An) and R2(B1,B2,...,Bm) is a
s5 jones cs zobel 1 cs270 zobel relation R3(C1,C2,...,Cn) such that
dom(Ci)= dom(Ai) −dom (Bi) for 1 ≤ i ≤ n
s6 peters ca kahn 3 cs270 woods
cs280 capon

The result R1 − R2 is a relation that includes

Lecturers π(lecturer)TEACH lecturer
kahn all tuples that are in R1 and not in R2
Tutors π(tutor)STUDENT bush
zobel
The resulting relation might have the same
Lecturers ∩ Tutors attribute names as the first or the second
relation

Retrieve all staff that lecture but don’t Outer Join Operation
tutor

In an equi-join, tuples without a ‘match’ are
STUDENT TEACH
studno name hons tutor year courseno lecturer eliminated
cs250 lindsey
s1 jones ca bush 2
Outer join keeps all tuples in R1 or R2 or both
s2 brown cis kahn 2 cs250 capon
s3 smith cs goble 2 cs260 kahn in the result, padding with nulls
s4 bloggs ca goble 1 cs260 bush
cs270 zobel

Left outer join R1 R2
s5 jones cs zobel 1
s6 peters ca kahn 3 cs270 woods
keeps every tuple in R1
cs280 capon
select * from R1, R2 where R1.a = R2.a (+)

Right outer join R1 R2
Lecturers π(lecturer)TEACH lecturer
keeps every tuple in R2
lindsey
Tutors π(tutor)STUDENT capon

select * from R1, R2 where R1.a (+) = R2.a
woods
Double outer join R1 R2
Lecturers -Tutors
keeps every tuple in R1 and R2

select * from R1, R2 where R1.a (+) = R2.a (+)

Outer Join Operator Outer Join Operator

STUDENT STAFF
STUDENT STAFF
studno name hons tutor year lecturer roomno
studno name hons tutor year lecturer roomno
s1 jones ca bush 2 kahn IT206
s1 jones ca bush 2 kahn IT206 s2 brown cis kahn 2 bush 2.26
s2 brown cis kahn 2 bush 2.26 s3 smith cs goble 2 goble 2.82
s3 smith cs goble 2 goble 2.82 s4 bloggs ca null 1 zobel 2.34
s4 bloggs ca null 1 zobel 2.34 s5 jones cs zobel 1 watson IT212
s5 jones cs zobel 1 watson IT212 s6 peters ca null 3 woods IT204
s6 peters ca null 3 woods IT204 capon A14
capon A14 lindsey 2.10
lindsey 2.10 select * from student, staff
barringer 2.125
select * from student, staff barringer 2.125 where tutor = lecturer (+)
where tutor = lecturer studno name hons tutor year lecturer roomno
studno name hons tutor year lecturer roomno s1 jones ca bush 2 bush 2.26
s1 jones ca bush 2 bush 2.26 s2 brown cis kahn 2 kahn IT206
s2 brown cis kahn 2 kahn IT206 s3 smith cs goble 2 goble 2.82
s3 smith cs goble 2 goble 2.82 s5 jones cs zobel 1 zobel 2.34
s5 jones cs zobel 1 zobel 2.34 s4 bloggs ca null 1 null null
s6 peters ca null 3 null null

6
 Outer Self Join
π (lecturer, (staff (appraiser = lecturer) staff) Outer Union
roomno,appraiser, approom)
Takes the union of tuples from two relations
STAFF lecturer roomno appraiser approom that are not union compatible
lecturer roomno appraiser kahn IT206 watson IT212
kahn IT206 watson bush 2.26 capon A14
The two relations, R1 and R2, are partially
bush 2.26 capon goble 2.82 capon A14 compatible—only some of their attributes are
goble 2.82 capon zobel 2.34 watson IT212
zobel 2.34 watson watson IT212 barringer 2.125 union compatible
watson IT212 barringer woods IT204 barringer 2.125
woods IT204 barringer capon A14 watson IT212
The attributes that are not union compatible
capon A14 watson lindsey 2.10 woods IT204
lindsey 2.10 woods
from either relation are kept in the result and
barringer 2.125 null null
barringer 2.125 null tuples without values for these attributes are
padded with nulls
select e.lecturer, e.roomno, m.lecturer appraiser, m.roomno approom
from staff e, staff m
where e.appraiser = m.lecturer (+)

Ordering results Completeness of Relational Algebra

select *

Five fundamental operations
from enrol, student
σ π X ∪ —
where labmark is not null and
Additional operators are defined as
combination of two or more of the basic
student.studno = enrol.studno operations,
order by
e.g.
R1 ∩ R2 = R1 ∪ R2 — ((R1 — R2) ∪ (R2—R1)
hons, courseno, name R1 <condition>R1 = σ<condition>(R1 X R2)

default is ascending

÷ Division Operation ÷ Division Operation Example

Definition: Retrieve the studnos of students who are enrolled
The division of two relations R1(A1,A2,...,An) on all the courses that Capon lectures on
with cardinality i and R2(B1,B2,...,Bm) with
Small_ENROL ÷ Capon_TEACH
cardinality j is a relation R3 with degree k=n-m,
cardinality i*j and attributes Small_ENROL
(A1,A2,...,An,B1,B2,...,Bm) studno courseno
s1 cs250
that satisfy the division condition s1 cs260
Capon_TEACH result
s1 cs280

The principle algebraic counterpart of queries s2 cs250
courseno s1
cs250
∀ s2 cs270 s4
that involve the universal quantifier s3 cs270 cs280

Relational languages do not express relational s4 cs280
s4 cs250
division s6 cs250

7
 Aggregation Functions Aggregation Functions in SQL

Aggregation functions on collections of data values:
ENROL select studno, count(*),
average, minimum, maximum, sum, count stud course lab exam avg(labmark)

Group tuples by value of an attribute and apply no no mark mark
s1 cs250 65 52 from enrol
aggregate function independently to each group of tuples s1 cs260 80 75 group by studno
s1 cs270 47 34
s2 cs250 67 55
ENROL <grouping attributes> ƒ <function list> (relation name) s2 cs270 65 71
stud course lab exam s3 cs270 49 50
no no mark mark studno ƒ COUNT courseno (ENROL)
s4 cs280 50 51
studno count avg
s1 cs250 65 52 s5 cs250 0 3 s1 3 64
s1 cs260 80 75 s6 cs250 2 7 s2 2 66
s1 cs270 47 34 studno count
s3 1 49
s1 3
s2 cs250 67 55 s4 1 50
s2 2 s5 1 0
s2 cs270 65 71
s3 1 s6 1 2
s3 cs270 49 50
s4 1
s4 cs280 50 51 s5 1
s5 cs250 0 3 s6 1
s6 cs250 2 7

Aggregation Functions in SQL Nested Subqueries

ENROL
stud course lab exam
Complete select queries within a where
no no mark mark
select studno, count(*),
avg(labmark) clause of another outer query
s1 cs250 65 52
s1 cs260 80 75 from enrol
Creates an intermediate result
s1 cs270 47 34
s2 cs250 67 55 group by studno
No limit to the number of levels of nesting
s2 cs270 65 71 having count(*) >= 2 List all students with the same tutor as bloggs
s3 cs270 49 50
s4 cs280 50 51
s5 cs250 0 3 select studno, name, tutor
s6 cs250 2 7
from student studno name tutor
studno count avg s3 smith goble
s1 3 64 where tutor =(select tutor s4 bloggs goble
s2 2 66
from student
where name = ‘bloggs’)

Union compatibility in nested

Nested Subqueries subqueries
select distinct name select distinct studno
from student from enrol
where studno in
where (courseno, exammark) in
(select studno
(select courseno, exammark
from enrol, teach, year
where from student s, enrol e
year.yeartutor = teach.lecturer and where s.name = ‘bloggs’ and
teach.courseno = enrol.courseno) e.studno = s.studentno);

8
Nested subqueries set comparison
Subqueries
operators
May be used in these situations:

Outer query qualifier includes a value v
compared with a bag of values V generated
to define the set of rows to be inserted in the
from a subquery target table of an insert, create table
or copy command

Comparison v with V evaluates TRUE

to define one or more values to be assigned
v in V if v is one of the elements V
to existing rows in an update statement
v = any V if v equal to some value in V
v > any V if v > some value in V (same for <)
to provide values for comparison in where,
v > all V if v greater than all the values in V having and start with clauses in
(same for <) select, update and delete commands

Correlated subqueries Exists and correlated sub queries

Exists is usually used to check whether the result of a

A condition in the where clause of a nested correlated nested query is empty
query references some attribute of a relation
Exists (Q) returns TRUE if there is at least one tuple in
declared in the outer (parent) query the results query Q and FALSE otherwise

The nested query is evaluated once for each select name

tuple in the outer (parent) query from student
where exists (select *
select name from enrol, teach
from student where student.studno=enrol.studno and
where 3 > (select count (*) enrol.courseno = teach.courseno and
from enrol teach.lecturer = ‘Capon’)
where student.studno=enrol.studno)
Retrieve the names of students who have registered for
at least one course taught by Capon

Exists and correlated sub queries

Conclusions

Not Exists (Q) returns FALSE if there is at
least one tuple in the results query Q and
The only logical structure is that of a relation
TRUE otherwise
Constraints are formally defined on the
select name concept of domain and key
from student
Operations deal with entire relations rather
where not exists (select * than single record at a time
from enrol

Operations are formally defined and can be
where student.studno=enrol.studno)
combined in a declarative language to
Retrieve the names of students who have no implement user queries on the database
enrolments on courses

9
 Conclusions on SQL

Retrieval:
many ways to achieve the same
result—though different performance costs

Comprehensive and powerful facilities

Non-procedural

Can’t have recursive queries

Limitations
are overcome by use of a high-
level procedural language that permits
embedded SQL statements

10