CS 461: Database Systems

Relational Algebra and SQL

supplementary material:
“Database Management Systems” Sec. 4.1, 4.2, 5.1-5.6
class notes

code: Sailors.sql, Query.sql

Animals.sql, Query2.sql

Julia Stoyanovich ([email protected])

 Game of Thrones
Characters Episodes Appearances
name house season num (tle viewers name season num

Eddard Stark 1 1 Winter is Coming 2.2 M Eddard 1 1

Jon Snow Stark 1 2 The Kingsroad 2.2 M Eddard 1 2

Tyrion Lannister Jon Snow 1 2

2 1 The North Remembers 3.9 M
Jon Snow 2 1
Daenerys Targaryen 2 2 The Night Lands 3.8 M
Jon Snow 2 2
Tyrion 1 1
select E.title, C.name, C.house
Tyrion 1 2
from Characters C, Episodes E, Appearances A
where C.house = ‘Targaryen’ Tyrion 2 2
and A.name = C.name Daenerys 1 2
and E.season = A.season
and E.num = A.num
Daenerys 2 1

What does this query compute?

How does it compute its results?

2
 Outline

• Part 1: Relational algebra

• Part 2: SQL (structured query language)

• Operations on a single relation

• Combining relations

• Aggregation

• Additional SQL features

3
 Select-project-join SQL queries
select <attribute list>
required from <relation list>
where <condition>
optional and <condition>
and ...

select E.title, C.name, C.house

from Characters C, Episodes E, Appearances A
where C.house = ‘Targaryen’
and A.name = C.name
and E.season = A.season
and E.num = A.num

• SPJ queries are the basic, and the most common, kind of a SQL query

• They read from at least one relation

• They return exactly one relation, which may be empty

• Because inputs and outputs are relations, SPJ queries are fully compositional -
this is called closure
4
 Game of Thrones
Characters Episodes Appearances
name house season num (tle viewers name season num

Eddard Stark 1 1 Winter is Coming 2.2 M Eddard 1 1

Jon Snow Stark 1 2 The Kingsroad 2.2 M Eddard 1 2

Tyrion Lannister Jon Snow 1 2

2 1 The North Remembers 3.9 M
Jon Snow 2 1
Daenerys Targaryen 2 2 The Night Lands 3.8 M
Jon Snow 2 2
select E.title, C.name, C.house Tyrion 1 1
from Characters C, Episodes E, Appearances A
Tyrion 1 2
where C.house = ‘Targaryen’
and A.name = C.name Tyrion 2 2
and E.season = A.season Daenerys 1 2
and E.num = A.num
Daenerys 2 1

Result (tle name house

The Kingsroad Daenerys Targaryen
The North Remembers Daenerys Targaryen

5
 Enter relational algebra

• SQL queries are compiled into relational algebra statement

• Formally: the data manipulation aspect of the relational model.

Takes relations as input, produces relations as output.

• Practically: a programming language for relational databases

• simpler and less powerful than a general programming language

• easier to learn (for us)

• easier to make efficient (for the database management system)

• Relational algebra is an algebra: relation variables / constants are

operands, operators act on such variables and constants

6
 What is an algebra?
• A system consisting of operators and operands

• We are all familiar with the algebra of arithmetic: operators

are + − ×, operands are constants, like 42, or variables, like x

• Expressions are made up of operators, operands, optionally

grouped by parentheses, e.g., (x + 3) / (y − 1)

• In relational algebra:

• operands are variables that stand for relations

• constants stand for finite relations (think: a particular set of

tuples)

• let’s look at operators

7
 Relational algebra operations
1. The usual set operations: union ∪ , intersection ∩, set
difference ∖ , but applied to relations (sets of tuples)
2. Operations that remove parts of a relation

• selection removes rows (tuples)

• projection removes columns (attributes)

3. Operations that combine tuples of two relations

• Cartesian product - pairs up tuples in two relations in

all possible ways

• join - selectively pairs up tuples from two relations

4. A renaming operation changes relation schema, re-
assigning either relation name or names of attributes
8
 Set operations on relations
Definition: Relations R and S are union-compatible if their
schemas define attributes with the same (or compatible)
domains.

Set operations can only be applied to union-compatible relations.

R R∪S R/S
id name age id name age id name age
1 Ann 18 1 Ann 18 2 Jane 22
2 Jane 22 2 Jane 22
3 Mike 21
S 4 Dave 27 S/R
id name age id name age
1 Ann 18 R∩S 3 Mike 21
3 Mike 21 id name age 4 Dave 27
4 Dave 27 1 Ann 18

Note: (1, Ann, 18) appears only once in the result of R ∪S

9
 Selection
The selection operator, applied to relation R, produces
a new relation with a subsets of R’s tuples. Tuples in σ C (R)
the new relation are those that satisfy some condition c.

Episodes σ viewers>3M Episodes

season num (tle viewers season num (tle viewers
1 1 Winter is Coming 2.2 M 2 1 The North Remembers 3.9 M
1 2 The Kingsroad 2.2 M 2 2 The Night Lands 3.8 M
2 1 The North Remembers 3.9 M
2 2 The Night Lands 3.8 M

Note: σ C (R) has at most as many rows as R

10
 Projection
The projection operator, applied to relation π A1 , A2 ,..., An (R)
R, produces a new relation with a subsets
of R’s attributes.

Episodes π season, title Episodes π season Episodes

season num (tle viewers season (tle season
1 1 Winter is Coming 2.2 M 1 Winter is Coming 1
1 2 The Kingsroad 2.2 M 1 The Kingsroad 2
2 1 The North Remembers 3.9 M 2 The North Remembers
2 2 The Night Lands 3.8 M 2 The Night Lands

Note: π A1 , A2 ,..., An (R) has at most as many rows as R

Why not exactly as many?
11
 Cartesian product
The Cartesian product (or cross product) of two
relations R and S is the set of pairs, formed by choosing R×S
the first element from R and the second element from S.
Characters
name house

Tyrion Lannister
Characters × Episodes
Daenerys Targaryen name house season num (tle
Tyrion Lannister 1 1 Winter is Coming
Episodes Tyrion Lannister 1 2 The Kingsroad
season num (tle Daenery Targaryen 1 1 Winter is Coming
sDaenery Targaryen 1 2 The Kingsroad
1 1 Winter is Coming
s
1 2 The Kingsroad

Note: there are exactly | R | ∗ | S | tuples in R × S

12
 Join
The join of two relations R and S is the set of pairs, formed
by choosing the first element from R and the second R ⨝C S
element from S, such that the corresponding tuples in R
and S meet some condition c .
Characters
name house

Tyrion Lannister

Daenerys Targaryen Characters >< name Appearances

name house name season num
Appearances Tyrion Lannister Tyrion 1 1
name season num Tyrion Lannister Tyrion 2 2
Jon Snow 2 1 Daenerys Targaryen Daenery 1 2
Tyrion 1 1 s
Tyrion 2 2
Daenerys 1 2
Note: there are at most | R |∗| S | tuples in R ⨝C S
13
Join vs. Cartesian product
Conceptually, to compute R ⨝C S
1. compute a Cartesian product R ⨉ S

2. then compute a selection σC (R ⨉ S) using the

join condition
R ⨝C S = σC (R ⨉S)
R ⨝R.age < S.age S = σ R.age < S.age (R ⨉S)
R.id R.name R.age S.id S.name S.age
1 Ann 18 3 Mike 21
1 Ann 18 4 Dave 27

2 Jane 22 3 Mike 21

2 Jane 22 4 Dave 27

14
 Natural join
The natural join of two relations R and S is a shorthand notation for a
join with the condition: the pair up tuples from R and S join if they R⨝S
agree on the values of the common attributes.
R S
sid name gpa sid did cid term grade
1111 Joe 3.2
1111 1 210 Fall 2012 A
2222 Ann 4.0
2222 1 220 Winter 2013
3333 Mike 3.5

R⨝S
R.sid R.name R.gpa S.sid S.did S.cid S.term S.grade
1111 Joe 3.2 1111 1 210 Fall 2012 A
2222 Ann 4.0 2222 1 220 Winter 2013

Note: there are at most | R | ∗ | S | tuples in R ⨝C S

15
 Renaming
Sometimes it is necessary to rename a
relation, or columns in the relation. For this,
we use the renaming operator.

Boats (bid, name, color) Vessels (vid, handle, shade)

ρVessels(vid, handle, shade) (Boats)

Boats (bid, name, color) Barges (bid, name, color)

ρBarges (Boats)
16
Relational algebra operations (recap)
1. The usual set operations: union ∪ , intersection ∩, set
difference ∖ , but applied to relations (sets of tuples)
2. Operations that remove parts of a relation

• selection removes rows (tuples)

• projection removes columns (attributes)

3. Operations that combine tuples of two relations

• Cartesian product - pairs tuples in two relations in all

possible ways

• join - selectively pairs tuples from two relations

4. A renaming operation changes relation schema, re-
assigning either relation name or names of attributes
17
 Behold the simplicity!

• A limited set of operations

• All operations take relations as input and produce

relations as output

• No pointers (i.e., no pointer chasing)

A first-order logic - based formalism, aimed specifically

at efficiently processing large datasets in bulk

18
 Combining operations into queries
What are the names of students whose GPA is at least 3.5?

Students sid name gpa

1111 Joe 3.2
2222 Ann 4.0
3333 Mike 3.5

1. Select those Students tuples that have gpa > 3.5.

2. Get (project) the value of the name column for these tuples.

πname ( σgpa>3.5 (Students)) name

Ann
Mike

19
 Combining operations into queries
What are the names of students who got an A in any course?

Students Enrollment
sid name gpa sid did cid term grade
1111 Joe 3.2 1111 1 210 Fall 2015 A
2222 Ann 4.0
2222 1 220 Winter 2016
3333 Mike 3.5

1. Join Students with Enrollment (natural join)

2. Select only those tuples where grade = ‘A’
3. Project the value of the name column for the resulting tuples

πname ( σgrade=’A’ ( Students ⨝ Enrollment)) name

Joe

20
 Combining operations into queries
What are the names of students who got an A in any course?

Students Enrollment
sid name gpa sid did cid term grade
1111 Joe 3.2 1111 1 210 Fall 2015 A
2222 Ann 4.0
2222 1 220 Winter 2016
3333 Mike 3.5

1. Select tuples in Enrollment where grade = ‘A’

2. Join these tuples with Students (natural join)
3. Project the value of the name column for the resulting tuples

πname ( Students ⨝ ( σgrade=’A’ Enrollment)) name

Joe

We compute the same result by executing operators in a different order!

21
 Examples
Sailors (sid, name, rating, age) Boats (bid, name, color) Reserves (sid, bid, day)
sid name ra(ng age bid name color sid bid day
1 DusQn 7 45 101 Interlake blue 1 101 10/10/12
2 Rusty 10 35 102 Interlake red 1 102 10/10/12
3 HoraQo 5 35 103 Clipper green 1 101 10/7/12
4 Zorba 8 18 104 Marine red 2 102 11/9/12
5 Julius 25
2 102 7/11/12

List names of boats. πname ( Boats) 3 101 7/11/12

3 102 7/8/12
4 103 19/9/12
List ratings and ages sailors. πrating, age (Sailors)
List names of sailors who are over 21 years old.
πname ( σage>21 (Sailors))
List names of red boats. πname ( σcolor=red (Boats))

22
 Examples
Sailors (sid, name, rating, age) Boats (bid, name, color) Reserves (sid, bid, day)
sid name ra(ng age bid name color sid bid day
1 DusQn 7 45 101 Interlake blue 1 101 10/10/12
2 Rusty 10 35 102 Interlake red 1 102 10/10/12
3 HoraQo 5 35 103 Clipper green 1 101 10/7/12
4 Zorba 8 18 104 Marine red 2 102 11/9/12
5 Julius 25
2 102 7/11/12
3 101 7/11/12
3 102 7/8/12
List ids of boats named Interlake 4 103 19/9/12

List ids of boats reserved on 10/10/12

23
 Examples (solution)
Sailors (sid, name, rating, age) Boats (bid, name, color) Reserves (sid, bid, day)
sid name ra(ng age bid name color sid bid day
1 DusQn 7 45 101 Interlake blue 1 101 10/10/12
2 Rusty 10 35 102 Interlake red 1 102 10/10/12
3 HoraQo 5 35 103 Clipper green 1 101 10/7/12
4 Zorba 8 18 104 Marine red 2 102 11/9/12
5 Julius 25
2 102 7/11/12
3 101 7/11/12
3 102 7/8/12
List ids of boats named Interlake 4 103 19/9/12

π bid (σ name=Interlake (Boats))

List ids of boats reserved on 10/10/12

π bid (σ day=10/10/12 (Reserves))

24
 Examples
Sailors (sid, name, rating, age) Boats (bid, name, color) Reserves (sid, bid, day)
sid name ra(ng age bid name color sid bid day
1 DusQn 7 45 101 Interlake blue 1 101 10/10/12
2 Rusty 10 35 102 Interlake red 1 102 10/10/12
3 HoraQo 5 35 103 Clipper green 1 101 10/7/12
4 Zorba 8 18 104 Marine red 2 102 11/9/12
5 Julius 25
2 102 7/11/12
3 101 7/11/12
List ids of sailors who reserved boat 102
3 102 7/8/12
πsid ( σbid=102 Reserves) 4 103 19/9/12

List names of sailors who reserved boat 102

πname (Sailors ⨝ ( σbid=102 Reserves)) both are correct!

which is better?
πname (σbid=102 (Sailors ⨝ Reserves))

25
 Examples
Sailors (sid, name, rating, age) Boats (bid, name, color) Reserves (sid, bid, day)
sid name ra(ng age bid name color sid bid day
1 DusQn 7 45 101 Interlake blue 1 101 10/10/12
2 Rusty 10 35 102 Interlake red 1 102 10/10/12
3 HoraQo 5 35 103 Clipper green 1 101 10/7/12
4 Zorba 8 18 104 Marine red 2 102 11/9/12
5 Julius 25
2 102 7/11/12
3 101 7/11/12
List names of sailors who reserved the red
3 102 7/8/12
Interlake. 4 103 19/9/12
πSailors.name (
Sailors ⨝ (
( σname=Interlake and color=red Boats) ⨝ Reserves))

any other way to do this?

26
 Examples
Sailors (sid, name, rating, age) Boats (bid, name, color) Reserves (sid, bid, day)
sid name ra(ng age bid name color sid bid day
1 DusQn 7 45 101 Interlake blue 1 101 10/10/12
2 Rusty 10 35 102 Interlake red 1 102 10/10/12
3 HoraQo 5 35 103 Clipper green 1 101 10/7/12
4 Zorba 8 18 104 Marine red 2 102 11/9/12
5 Julius 25
2 102 7/11/12
3 101 7/11/12
List names of boats that were reserved by 3 102 7/8/12
Horatio. 4 103 19/9/12

πBoats.name (
(σSailors.name=Horatio Sailors) ⨝
( Boats ⨝ Reserves))
any other way to do this?

27
 Examples
Sailors (sid, name, rating, age) Boats (bid, name, color) Reserves (sid, bid, day)
sid name ra(ng age bid name color sid bid day
1 DusQn 7 45 101 Interlake blue 1 101 10/10/12
2 Rusty 10 35 102 Interlake red 1 102 10/10/12
3 HoraQo 5 35 103 Clipper green 1 101 10/7/12
4 Zorba 8 18 104 Marine red 2 102 11/9/12
5 Julius 25
2 102 7/11/12
3 101 7/11/12
List days on which some sailor with rating 3 102 7/8/12
higher than 7 was at sea 4 103 19/9/12

28
 Examples (solution)
Sailors (sid, name, rating, age) Boats (bid, name, color) Reserves (sid, bid, day)
sid name ra(ng age bid name color sid bid day
1 DusQn 7 45 101 Interlake blue 1 101 10/10/12
2 Rusty 10 35 102 Interlake red 1 102 10/10/12
3 HoraQo 5 35 103 Clipper green 1 101 10/7/12
4 Zorba 8 18 104 Marine red 2 102 11/9/12
5 Julius 25
2 102 7/11/12
3 101 7/11/12
List days on which some sailor with rating 3 102 7/8/12
higher than 7 was at sea 4 103 19/9/12

πday ((σrating>7 Sailors) ⨝ Reserves)

29
 Examples
Sailors (sid, name, rating, age) Boats (bid, name, color) Reserves (sid, bid, day)
sid name ra(ng age bid name color sid bid day
1 DusQn 7 45 101 Interlake blue 1 101 10/10/12
2 Rusty 10 35 102 Interlake red 1 102 10/10/12
3 HoraQo 5 35 103 Clipper green 1 101 10/7/12
4 Zorba 8 18 104 Marine red 2 102 11/9/12
5 Julius 25
2 102 7/11/12
3 101 7/11/12
List names and colors of boats that were 3 102 7/8/12
reserved by Zorba 4 103 19/9/12

30
 Examples (solution)
Sailors (sid, name, rating, age) Boats (bid, name, color) Reserves (sid, bid, day)
sid name ra(ng age bid name color sid bid day
1 DusQn 7 45 101 Interlake blue 1 101 10/10/12
2 Rusty 10 35 102 Interlake red 1 102 10/10/12
3 HoraQo 5 35 103 Clipper green 1 101 10/7/12
4 Zorba 8 18 104 Marine red 2 102 11/9/12
5 Julius 25
2 102 7/11/12
3 101 7/11/12
List names and colors of boats that were 3 102 7/8/12
reserved by Zorba 4 103 19/9/12

πBoats.name, Boats.color (
(σname=Zorba Sailors) ⨝ Reserves) ⨝ Boats))

31
 Renaming
Sometimes it is necessary to rename a relation, or
columns in the relation. For this, we use the renaming
operator.

Boats (bid, name, color) Vessels (vid, handle, shade)

ρVessels(vid, handle, shade) (Boats)

Boats (bid, name, color) Barges (bid, name, color)

ρBarges (Boats)
32
 A self-join
A self-join is a join that joins together tuples from two
copies of the same table

List all possible heterosexual couples (girl name, boy name),

where the boy is older than the girl.

People id name age gender

1 Ann 18 F

2 Jane 22 F
3 Mike 21 M
4 Dave 27 M

πGirls.name, Boys.name (
ρGirls ( σgender=F People ) ⨝Girls.age<Boys.age
ρBoys ( σgender=M People ) )

33
 Relational algebra - recap

• Is the data manipulation aspect of the relational

model

• Expressions are procedural, in that the order of

operators is specified explicitly

• Often several different relational algebra

expressions will give the same result (for all valid
instances!), but will have different performance
characteristics

• SQL can be seen as a declarative implementation of

relational algebra

34
 Outline

• Part 1: Relational algebra

• Part 2: SQL

• Operations on a single relation

• Combining relations

• Aggregation

• Additional SQL features

35
 SQL

• SQL (“seekwel”) stands for Structured Query

Language

• Made up of 2 parts:

• Data Definition Language (DDL) - used to

create or modify a relational schema

• Data Manipulation Language (DML) - used

to retrieve or modify data in a schema

• We call SQL statements queries

36
 SQL, relational algebra, do we need both?
Characters Episodes Appearances
name house season num (tle viewers name season num

Eddard Stark 1 1 Winter is Coming 2.2 M Eddard 1 1

Jon Snow Stark 1 2 The Kingsroad 2.2 M Eddard 1 2

Tyrion Lannister Jon Snow 1 2

2 1 The North Remembers 3.9 M
Jon Snow 2 1
Daenerys Targaryen 2 2 The Night Lands 3.8 M
Jon Snow 2 2
π title, name, house Tyrion 1 1
select E.title, C.name, C.house
Tyrion 1 2
from Characters C, Episodes E,
Appearances A >< 2 tuples Tyrion 2 2
where C.house = ‘Targaryen’ Daenerys 1 2
and A.name = C.name
2 tuples
and E.season = A.season σ house=Trgaryen Episodes
Daenerys 2 1
and E.num = A.num

10 tuples ><

Characters Appearances
37
 SQL, relational algebra, do we need both?
Characters Episodes Appearances
name house season num (tle viewers name season num

Eddard Stark 1 1 Winter is Coming 2.2 M Eddard 1 1

Jon Snow Stark 1 2 The Kingsroad 2.2 M Eddard 1 2

Tyrion Lannister Jon Snow 1 2

2 1 The North Remembers 3.9 M
Jon Snow 2 1
Daenerys Targaryen 2 2 The Night Lands 3.8 M
Jon Snow 2 2
π title, name, house Tyrion 1 1
select E.title, C.name, C.house
from Characters C, Episodes E, Tyrion 1 2
Appearances A
where E.season = A.season >< 2 tuples Tyrion 2 2
and E.num = A.num 2 tuples Daenerys 1 2
and A.name = C.name Daenerys 2 1
and C.house = ‘Targaryen’ >< Episodes

1 tuple σ house=Trgaryen

Characters Appearances
38
 SQL, relational algebra, do we need both?
Characters Episodes Appearances
name house season num (tle viewers name season num

Eddard Stark 1 1 Winter is Coming 2.2 M Eddard 1 1

Jon Snow Stark 1 2 The Kingsroad 2.2 M Eddard 1 2

Tyrion Lannister Jon Snow 1 2

2 1 The North Remembers 3.9 M
Jon Snow 2 1
Daenerys Targaryen 2 2 The Night Lands 3.8 M
Jon Snow 2 2
π title, name, house Tyrion 1 1
select E.title, C.name, C.house
from Characters C, Episodes E, Tyrion 1 2
Appearances A
where E.season = A.season
σ 2 tuples
join( A,C )∧ join( A,E )∧house=Trgaryen
Tyrion 2 2
and E.num = A.num Daenerys 1 2
and A.name = C.name Daenerys 2 1
and C.house = ‘Targaryen’ 160 tuples ×

40 tuples × Episodes

Characters Appearances
39
How do we pick a good rewriting?
• Good question! The answer: we don’t!

• Rewriting SQL queries into relational algebra statements (or

some similar representation) is the job of the query optimizer

• We, people, do what we do well: specify what we want to

compute using SQL

• The system does what it does well: decides on an efficient

query evaluation procedure - how to compute the result

• I got one word for you - declarativeness!

40
 Architecture of a typical DBMS

Application Query evaluation parser, compiler,

engine optimizer, evaluator

Concurrency Recovery
Storage manager
control manager

transaction manager
index/record manager,
lock manager
buffer manager,
Storage disk space manager

41
 Overview of query optimization

Given a SQL query, there may be different execution plans

that produce the same result but that have different
performance characteristics

• Goal of query optimization: find an efficient query execution

plan for a given query

• Query execution plan is represented by a tree of relational

algebra operators, annotated with a choice of an algorithm for
each operator

• The optimizer also decides how to access data in a particular

relation, i.e., which access path to use

42
 Operations on a single relation

R
id name age gender
1 Ann 18 F
Given a relation R, return a new
2 Jane 22 F
relation R’ that contains a sub-
3 Mike 21 M
set of the rows / columns from R
4 Dave 27 M

43
 Selection in SQL
σ C (R) select *
from R
where C

R σ age≥21 (R) σ age≥21 AND gender=' M ' (R)

select *
id name age gender select *
from R
1 Ann 18 F from R
where age >= 21
where age >= 21
2 Jane 22 F and gender =’M’
3 Mike 21 M
id name age gender id name age gender
4 Dave 27 M
2 Jane 22 F 3 Mike 21 M
3 Mike 21 M 4 Dave 27 M
4 Dave 27 M

44
 Projection in SQL
π A1 , A2 ,..., An (R) select A1, A2, ..., An
from R

R π id, name (R) π gender (R)

id name age gender gender
select id, name
1 Ann 18 F F
from R
2 Jane 22 F M
3 Mike 21 M id name
1 Ann select gender
4 Dave 27 M
2 Jane from R
3 Mike gender
4 Dave F
45
F
M
M
SQL queries compute bags, not sets!
• A bag (aka a multi-set) is an unordered collection with duplicates

• This is unlike a set, which is also unordered, but there are no

duplicates

• Relational model, relational algebra work with sets

• SQL works with bags

• Why bags: efficiency of implementation is the main reason

R select gender select distinct gender
from R from R
id name age gender gender gender
1 Ann 18 F F F
2 Jane 22 F F M
3 Mike 21 M M
4 Dave 27 M M

We use distinct to covert from a bag to a set

46
 Combining selection and projection

π A1 , A2 ,..., An (σ C (R)) select A1, A2, ..., An

from R
where C

What are the names of students whose GPA is at least 3.5?

Students sid name gpa

1111 Joe 3.2
2222 Ann 4.0
3333 Mike 3.5

select name
π name (σ gpa≥3.5 (Students)) from Students
where gpa >= 3.5

name
Ann
Mike

47
 Examples
Sailors (sid, name, rating, age) Boats (bid, name, color) Reserves (sid, bid, day)
sid name ra(ng age bid name color sid bid day
1 DusQn 7 45 101 Interlake blue 1 101 10/10/12
2 Rusty 10 35 102 Interlake red 1 102 10/10/12
3 HoraQo 5 35 103 Clipper green 1 101 10/7/12
4 Zorba 8 18 104 Marine red 2 102 11/9/12
5 Julius 25 2 102 7/11/12
3 101 7/11/12
(a) List names of boats. 3 102 7/8/12
4 103 19/9/12
(b) List ratings and ages sailors.

(c) List names of sailors who are over 21 years old.

(d) List names of red boats.

(e) List ids of boats that have ever been reserved.

48
 String expressions
Sailors (sid, name, rating, age) Boats (bid, name, color) Reserves (sid, bid, day)

Find name, age, 2*rating of all sailors whose name starts with A or
contains ‘ust’ as a substring starting from position 2
Q1 in Query.sql
select name, age, rating * 2 as twice_the_rating
from Sailors
where name like 'A%' or name like '_ust%';

as is a way to name a field in the result

% denotes 0 or more arbitrary characters

_ denotes exactly 1 arbitrary character

like is used for string matching
49
 Expressions and strings: examples
Sailors (sid, name, rating, age) Boats (bid, name, color) Reserves (sid, bid, day)

Find names, sids of sailors whose rating * 5 is less than

Q2 their age
select name, sid
from Sailors
where rating * 5 < age

Find names, sids of sailors whose names contain the

letter u
Q3
select name, sid
from Sailors
where name like '%u%'

50
 Null values
• null stands for a missing / unknown / inapplicable value

• SQL provides special comparison operators for null

(cannot use = < > !=)
select *
Q4 from Sailors
where rating is not null;

Sailors (sid, name, rating, age) sid name ra(ng age

sid name ra(ng age 1 DusQn 7 45

1 DusQn 7 45 2 Rusty 10 35

2 Rusty 10 35 3 HoraQo 5 35

3 HoraQo 5 35 4 Zorba 8 18

4 Zorba 8 18 select *
5 Julius null 25 Q5 from Sailors
where rating is null;
sid name ra(ng age

5 Julius null 25

51
 Null values (II)
• null stands for a missing / unknown / inapplicable value

• SQL provides special comparison operators for null

(cannot use = < > !=)
select sid, rating * 0
Sailors (sid, name, rating, age) Q6 from Sailors;
sid ra(ng
sid name ra(ng age
1 0
1 DusQn 7 45
2 Rusty 10 35 2 0

3 HoraQo 5 35 3 0
4 Zorba 8 18 4 0
5 Julius null 25 5 null

The value of rating * 0 for sid = 5 is null, not 0!

52
 Renaming attributes
select sid as sailor_id, rating / 10 as normalized_rating
Q7 from Sailors;

the as keyword is optional

select sid sailor_id, rating / 10 normalized_rating

from Sailors;

sailor_id normalized_ra(ng
1 0.7
2 1

3 0.5

4 0.8

5 null

53
 Renaming relations
Q8 select S.sid, S.rating / 10 as normalized_rating
from Sailors S;

sid normalized_ra(ng
1 0.7
2 1

3 0.5

4 0.8

5 null

54
 Sorting results
• Sets and bags are unordered collections

• Sometimes we want to see results in sorted

order, an order by clause allows us to do this

select Ai, Aj, ..., An

from R
where C
order by Ak [desc], Al [desc], ..., Am [desc]

1. compute the result of π Ai , A j ,..., An (σ C (R))

2. sort the result by Ak, break ties by Al, ...., break ties by Am

3. the optional desc keyword after each column specifies descending (higher
to lower) sort order; the default order is ascending (lower to higher)

55
 Sorting results: example
Retrieve all reservations sorted by sid, Retrieve all reservations sorted by
with ties broken by bid sid (descending), with ties broken
by bid (ascending)
select * select *
Q9 from Reserves Q10 from Reserves
order by sid, bid order by sid desc, bid

sid bid day sid bid day

1 101 10/10/12 4 103 19/9/12
1 101 10/7/12 3 101 7/11/12
1 102 10/10/12 3 102 7/8/12
2 102 11/9/12 2 102 11/9/12
2 102 7/11/12 2 102 7/11/12
3 101 7/11/12 1 101 10/10/12
3 102 7/8/12 1 101 10/7/12
4 103 19/9/12 1 102 10/10/12

56
 Examples
Sailors (sid, name, rating, age) Boats (bid, name, color) Reserves (sid, bid, day)
bid name color sid bid day
sid name ra(ng age
101 Interlake blue 1 101 10/10/12
1 DusQn 7 45
102 Interlake red 1 102 10/10/12
2 Rusty 10 35
103 Clipper green 1 101 10/7/12
3 HoraQo 5 35
104 Marine red 2 102 11/9/12
4 Zorba 8 18
5 Julius 25 2 102 7/11/12
3 101 7/11/12
(a) Find the color of a boat named clipper. 3 102 7/8/12
4 103 19/9/12

(b) Find all sailors who are 35 or older, sort them by name.

(c) List the names of red boats, sorted by boat id.

(d) Find all boats that are either red or called Interlake.

57
 Combining relations
R
id name age gender
1 Ann 18 F Informally: given relations R, S,
2 Jane 22 F return a new relation T that
3 Mike 21 M
contains some combination of
4 Dave 27 M
tuples from R, S

S
id major gpa
1 Math 3.7

2 CS 3.8
3 English 3.8
4 Philosophy 3.2

58
 Cartesian product
List all pairs of Girls (G) and Boys (B)
G
id name age
select *
from G, B G×B
1 Ann 18 G.id G.name G.age B.id B.name B.age
2 Jane 22 1 Ann 18 3 Mike 21

B 1 Ann 18 4 Dave 27

id name age 2 Jane 22 3 Mike 21

3 Mike 21 2 Jane 22 4 Dave 27

4 Dave 27

List all pairs of Girls (G) and Boys (B) that involve Ann
select *
from G, B σ G .name=' Ann' (G × B)
where G.name = ‘Ann’

G.id G.name G.age B.id B.name B.age

1 Ann 18 3 Mike 21
1 Ann 18 4 Dave 27

59
 Cartesian product
List all pairs of Girls (G) and Boys (B) where the girl is younger than the boy
select *
G
σG.age<B.age (G ⨉B)
from G, B
id name age where G.age < B.age
1 Ann 18

2 Jane 22
G.id G.name G.age B.id B.name B.age
1 Ann 18 3 Mike 21
B 1 Ann 18 4 Dave 27
id name age 2 Jane 22 4 Dave 27
3 Mike 21

4 Dave 27
or is this a join?

G ⨝G.age<B.age B = σG.age<B.age (G ⨉B)

yes, both are right, both are expressed by the same SQL query!

60
 Join
Students Enrollment
sid name gpa sid did cid term grade
1111 Joe 3.2
1111 1 210 Fall 2012 A
2222 Ann 4.0
2222 1 220 Winter 2013 null
3333 Mike 3.5

List the name, GPA and enrollment information for all students
who have been enrolled in a course.

Students ⨝ Enrollment
select S.name, S.gpa, E.did, E.cid, E.term
S.name S.gpa E.did E.cid E.term from Students S, Enrollment E
Joe 3.2 1 210 Fall 2012 where S.sid = E.sid
Ann 4 1 220 Winter 2013

Note that renaming relations is handy here

61
 Join
Students Enrollment
sid name gpa
sid did cid term grade
1111 Joe 3.2
1111 1 210 Fall 2012 A
2222 Ann 4.0
2222 1 220 Winter 2013 null
3333 Mike 3.5

List the name, GPA and enrollment information for all students who
have been enrolled in a course and have a grade for that course.

Students ⨝ σE.grade is not null (Enrollment)

σ E.grade is not null (Students ⨝ Enrollment)
select S.name, S.gpa, E.did, E.cid, E.term
S.sid E.did E.cid E.term from Students S, Enrollment E
1111 1 210 Fall 2012 where S.sid = E.sid
2222 1 220 Winter 2013 and E.grade is not null

Note that the same SQL query corresponds to both relational algebra
statements.

62
 SQL is declarative
Students ⨝ σE.grade is not null (Enrollment)
σ E.grade is not null (Students ⨝ Enrollment)
select S.name, S.gpa, E.did, E.cid, E.term
from Students S, Enrollment E
where S.sid = E.sid

• A relational algebra statement specifies the order of operations

• the first statement first selects and then joins, while the second first
joins and then selects

• in a sense, we have to specify both what to compute and how to

compute it

• Both relational algebra statements return the same result, but the
processing costs differ

• SQL queries are declarative: we only say what to compute not how

• This makes it easier for us, and leads to a more efficient query execution
plan (because the database engine knows best)

63
 More examples
Sailors (sid, name, rating, age) Boats (bid, name, color) Reserves (sid, bid, day)
bid name color sid bid day
sid name ra(ng age
101 Interlake blue 1 101 10/10/12
1 DusQn 7 45
102 Interlake red 1 102 10/10/12
2 Rusty 10 35
103 Clipper green 1 101 10/7/12
3 HoraQo 5 35
104 Marine red 2 102 11/9/12
4 Zorba 8 18
5 Julius null 25 2 102 7/11/12
3 101 7/11/12
3 102 7/8/12
4 103 19/9/12

Find names of sailors who have reserved boat 101

Q11 select
from
S.name
Sailors S, Reserves R
name

DusQn
where S.sid = R.sid
and R.bid = 101 DusQn

HoraQo

64
 More examples
Sailors (sid, name, rating, age) Boats (bid, name, color) Reserves (sid, bid, day)
bid name color sid bid day
sid name ra(ng age
101 Interlake blue 1 101 10/10/12
1 DusQn 7 45
102 Interlake red 1 102 10/10/12
2 Rusty 10 35
103 Clipper green 1 101 10/7/12
3 HoraQo 5 35
104 Marine red 2 102 11/9/12
4 Zorba 8 18
5 Julius null 25 2 102 7/11/12
3 101 7/11/12
3 102 7/8/12
4 103 19/9/12

Find ages of sailors who have reserved a red boat

select S.age
age
Q12 from Sailors S, Reserves R, Boats B 45
where S.sid = R.sid 35
and R.bid = B.bid
35
and B.color = ‘red’
35

65
 More examples
Sailors (sid, name, rating, age) Boats (bid, name, color) Reserves (sid, bid, day)
bid name color sid bid day
sid name ra(ng age
101 Interlake blue 1 101 10/10/12
1 DusQn 7 45
102 Interlake red 1 102 10/10/12
2 Rusty 10 35
103 Clipper green 1 101 10/7/12
3 HoraQo 5 35
104 Marine red 2 102 11/9/12
4 Zorba 8 18
5 Julius null 25 2 102 7/11/12
3 101 7/11/12
3 102 7/8/12
4 103 19/9/12

Find unique ages of sailors who have reserved a red boat

select distinct S.age age
Q13 from
where
Sailors S, Reserves R, Boats B
S.sid = R.sid
45

and R.bid = B.bid 35

and B.color = ‘red’

66
How do we read a SQL query?
select S.name, B.name, R.day
from Sailors S, Reserves R, Boats B
where S.sid = R.sid
and R.bid = B.bid
and B.color = ‘green’
order by S.name, R.day

• First, look at the from clause - which tables are being used

• Next, look at the where where clause - what are the selection and
join conditions

• Then, look at the select clause - which columns are used in the
result

• Finally, if required, think about the order in which results are

returned

To write queries, think in the same order: from, where,

select, order by.

67
 More examples
Sailors (sid, name, rating, age) Boats (bid, name, color) Reserves (sid, bid, day)
bid name color sid bid day
sid name ra(ng age
101 Interlake blue 1 101 10/10/12
1 DusQn 7 45
102 Interlake red 1 102 10/10/12
2 Rusty 10 35
103 Clipper green 1 101 10/7/12
3 HoraQo 5 35
104 Marine red 2 102 11/9/12
4 Zorba 8 18
5 Julius null 25 2 102 7/11/12
3 101 7/11/12
Find unique names and ratings of sailors who 3 102 7/8/12

have reserved a boat named Interlake and 4 103 19/9/12

whose name contains the letter ‘u’.

Q14*
Find days of reservations made for a red boat
by a sailor who is under 40 years old.

Q15*
68
 More examples
Sailors (sid, name, rating, age) Boats (bid, name, color) Reserves (sid, bid, day)
bid name color sid bid day
sid name ra(ng age
101 Interlake blue 1 101 10/10/12
1 DusQn 7 45
102 Interlake red 1 102 10/10/12
2 Rusty 10 35
103 Clipper green 1 101 10/7/12
3 HoraQo 5 35
104 Marine red 2 102 11/9/12
4 Zorba 8 18
5 Julius null 25 2 102 7/11/12
3 101 7/11/12
Find unique names and ratings of sailors who 3 102 7/8/12

have reserved a boat named Interlake and 4 103 19/9/12

whose name contains the letter ‘u’.

name ra(ng
select distinct S.name, S.rating
Q14 from Sailors S, Reserves R, Boats B DusQn 7
where S.sid = R.sid
Rusty 10
and R.bid = B.bid
and B.name = 'Interlake'
and S.name like '%u%'

69
 More examples
Sailors (sid, name, rating, age) Boats (bid, name, color) Reserves (sid, bid, day)
bid name color sid bid day
sid name ra(ng age
101 Interlake blue 1 101 10/10/12
1 DusQn 7 45
102 Interlake red 1 102 10/10/12
2 Rusty 10 35
103 Clipper green 1 101 10/7/12
3 HoraQo 5 35
104 Marine red 2 102 11/9/12
4 Zorba 8 18
5 Julius null 25 2 102 7/11/12
3 101 7/11/12
3 102 7/8/12

Find days of reservations made for a red boat 4 103 19/9/12

by a sailor who is under 40 years old.

day
Q15 select R.day
11/9/12
from Sailors S, Reserves R, Boats B
where S.sid = R.sid 7/11/12
and R.bid = B.bid 7/8/12
and B.color = 'red'
and S.age < 40

70
 Union, Intersect, Except
Find unique sids and names of sailors who have reserved a red
or a blue boat
select DISTINCT S.sid, S.name
Q16 from Sailors S, Boats B, Reserves R
where S.sid = R.sid and B.bid = R.bid
and (B.color = 'red' or B.color = 'blue')

Find unique sids and names of sailors who have reserved a

red and a blue boat
select DISTINCT S.sid, S.name
from Sailors S, Boats B1, Reserves R1,

Q17 Boats B2, Reserves R2

where S.sid = R1.sid and B1.bid = R1.bid
and S.sid = R2.sid and B2.bid = R2.bid
and B1.color = 'red'
and B2.color = 'blue'

Difficult to understand, is there an easier way?

71
 Union, Intersect, Except
Find sids of sailors who have reserved a red or a blue boat

Option 1 select DISTINCT S.sid

from Sailors S, Boats B, Reserves R
Q16 where S.sid = R.sid and B.bid = R.bid
and (B.color = 'red' or B.color = 'blue')

Option 2 select S.sid

from Sailors S, Boats B, Reserves R
Q18 where
and
S.sid = R.sid and B.bid = R.bid
B.color = 'red'
UNION
or select S.sid
from Sailors S, Boats B, Reserves R
where S.sid = R.sid and B.bid = R.bid
and B.color = 'blue';

no need for a DISTINCT

UNION returns a set

72
 Union, Intersect, Except
Find sids of sailors who have reserved a red and a blue boat
select DISTINCT S.sid
Option 1 from Sailors S, Boats B1, Reserves R1,
Boats B2, Reserves R2
Q17 where S.sid = R1.sid and B1.bid = R1.bid
and S.sid = R2.sid and B2.bid = R2.bid
and B1.color = 'red'
and B2.color = 'blue';

Option 2 select S.sid

from Sailors S, Boats B, Reserves R
Q19 where S.sid =
and B.color
R.sid and B.bid = R.bid
= 'red'
INTERSECT
and select S.sid
from Sailors S, Boats B, Reserves R
where S.sid = R.sid and B.bid = R.bid
and B.color = 'blue';

no need for a DISTINCT

INTERSECT returns a set

73
 Union, Intersect, Except
Find sids of sailors who have reserved a red but not a blue boat

Option 1: involves correlated subqueries, will see later

Option 2 select DISTINCT S.sid

from Sailors S, Boats B, Reserves R
Q20 where
and
S.sid = R.sid and
B.color = 'red'
B.bid = R.bid

EXCEPT
but not select
from
DISTINCT S.sid
Sailors S, Boats B, Reserves R
where S.sid = R.sid and B.bid = R.bid
and B.color = 'blue';

74
 To use set operators...
... we must make sure that the sets involved in the
UNION / INTERSECT / EXCEPT are union-
compatible, i.e., tuples have the same number of
fields, and fields are of compatible types.

select S.sid, S.name from Sailors S

UNION
select R.sid, R.day from Reserves R;

ERROR: UNION types character varying and date cannot be matched

LINE 3: select R.sid, R.day from Reserves R;
^

75
 More examples
Find unique bids of boats that have been reserved by
sailors with rating > 7

Q21 select
from
distinct R.bid
Reserves R, Sailors S
where R.sid = S.sid
and S.rating > 7

Find unique bids of boats that have been reserved by

sailors with rating > 7 or rating < 5
select DISTINCT R.bid select R.bid
from Reserves R, Sailors S from Reserves R, Sailors S
where
and
R.sid = S.sid
(S.rating > 7 or S.rating < 5)
where
and
R.sid = S.sid
S.rating > 7 Q23
UNION

Q22 select
from
R.bid
Reserves R, Sailors S
where R.sid = S.sid
and S.rating < 5

76
 More examples
Find bids of boats that have been reserved by sailors
with rating > 7 and rating < 5
Q24*

Find bids of boats that have been reserved by sailors

with rating > 7 but not rating < 5
Q25*

77
 More examples
Find bids of boats that have been reserved by sailors
with rating > 7 and rating < 5 select R.bid
from Reserves R, Sailors S
Q24 where R.sid = S.sid
and S.rating > 7
INTERSECT
select R.bid
from Reserves R, Sailors S
where R.sid = S.sid
and S.rating < 5

Find bids of boats that have been reserved by sailors

with rating > 7 but not rating < 5 select R.bid
from Reserves R, Sailors S

Q25 where
and
B.bid = R.bid
R.sid = S.sid
and S.rating > 7
EXCEPT
select R.bid
from Reserves R, Sailors S
where R.sid = S.sid
and S.rating < 5

78
 Nested queries

• Sometimes, it is necessary to embed (i.e., nest) a

query within a query (usually in the from or in the
where clause)

• This is needed when we have to refer to a value that

must itself be computed

• The embedded query is usually called a subquery

• You can nest queries within nested queries within

nested queries....

79
 Nested queries
Find names of sailors who have reserved boat 103
Option 1
select S.name
Q26 from Sailors S, Reserves R
where S.sid = R.sid
and R.bid = 103

Option 2 select S.name

from Sailors S
Q27 where S.sid in (select R.sid
from Reserves R
where R.bid = 103)

• subquery computes sids of sailors who reserved boat 103

• top-level query gets names of sailors whose sids are in this

(multi)set

80
 Nested queries: examples
Find names of sailors who have reserved a red boat

Option 1
Q28 select
from
S.name
Boats B, Reserves R, Sailors S
where B.bid = R.bid
and R.sid = S.sid
and B.color = 'red'

Option 2
select S.name
Q29 from Sailors S
where S.sid in (select R.sid
from Reserves R
where R.bid in (select B.bid
from Boats B
where B.color = 'red'))

81
Nested queries: more examples
Write two queries that compute this: one nested query and
one using join.

Find unique colors of boats that were reserved by sailors

with rating >7.
Q30*

Find unique names of boats that were reserved during the

month of October.
(to get reservations made during the month of October:
select * from Reserves where to_char(day, 'Mon') = 'Oct')
Q31*
82
Nested queries: more examples
Write two queries that compute this: one nested query
and one using join.

Find unique colors of boats that were reserved by sailors

with rating >7.

Q30.a Q30.b
select distinct B.color select distinct B.color
from Boats B, Reserves R, Sailors S from Boats B, Reserves R
where B.bid = R.bid where B.bid = R.bid
and R.sid = S.sid and R.sid in (select S.sid
and S.rating > 7 from Sailors S
where S.rating > 7)

83
Nested queries: more examples
Write two queries that compute this: one nested query and
one using join.

Find unique names of boats that were reserved during the

month of October.
(to get reservations made during the month of October:
select * from Reserves where to_char(day, 'Mon') = 'Oct')
select distinct B.name
Q31.a from Boats B, Reserves R
where B.bid = R.bid
and to_char(day, 'Mon') = 'Oct'

Q31.b select distinct B.name

from Boats B
where B.bid in (select R.bid
from Reserves R
where to_char(day, 'Mon') = 'Oct')

84
 Nested queries: examples
Find names of sailors who have never reserved a red boat

Q32 select S.name

from Sailors S
where S.sid not in (select R.sid
from Reserves R
where R.bid in (select B.bid
from Boats B
where B.color = 'red'))

What does this query compute?

Q33 select S.name
from Sailors S
where S.sid in (select R.sid
from Reserves R
where R.bid not in (select B.bid
from Boats B
where B.color = 'red'))

85
 Aggregation
• In addition to retrieving data, we can use SQL to
summarize / help analyze it.

• SQL supports operations for computing aggregate

values of any column A

• COUNT ([DISTINCT] A) - the number of (unique)

values in the A column

• SUM ([DISTINCT] A) - the sum of all (unique) values

in the A column

• AVG ([DISTINCT] A) - the average of all (unique)

values in the A column

• MIN (A) - the minimum value in the A column

• MAX (A) - the maximum value in the A column

86
 Aggregation: examples

Find the average age of sailors with a rating of 10

select AVG(age)
Q37 from Sailors
where rating = 10

Find the number of sailors

select COUNT(*)
Q38 from Sailors
think of * as shorthand for all columns

Find the number of sailors, their average age, and minimum

rating
select COUNT(*), AVG(age), MIN(rating)
Q39 from Sailors

87
Aggregation: more examples
Find the name and age of the oldest sailor

Take 1 (doesn’t work!) select name, MAX(age) as age

from Sailors

Rule 1: a query that uses aggregate operators must use

only aggregate operators in the select clause, unless the
query contains a group by clause (we’ll see this next).

Take 2 (works)

Q40 select S1.name, S1.age

from Sailors S1
where S1.age = (select max(S2.age) from Sailors S2)

this is also an example of a nested query in the where clause

88
Aggregation: more examples
Count the number of sailors
select COUNT(*)
from Sailors
Q38

Count the number of sailor ages

select COUNT(age)
from Sailors

Count the number of unique sailor ages

select COUNT(distinct age)
from Sailors

Count the number of sailor ages, and of unique sailor ages

select COUNT(distinct age) num_unique_ages, COUNT(age) num_ages
from Sailors
Q41
What about rating, rather than age?
89
Aggregation: more examples
Find the number of sailors who are older than the
oldest sailor with the rating of 10

Q42*

List sids, names of sailors with an above average rating

Q43*

90
Aggregation: more examples
Find the number of sailors who are older than the
oldest sailor with the rating of 10

Q42 select COUNT(*)

from Sailors S1
where S1.age > (select MAX(S2.age)
from Sailors S2
where S2.rating = 10)

List sids, names of sailors with an above average rating

Q43 select S1.sid, S1.name
from Sailors S1
where S1.rating > (select AVG(S2.rating)
from Sailors S2)

91
 Group-by and having clauses
Find the age of the youngest sailor for each rating level
select MIN(age) select MIN(age) select MIN(age)
from Sailors from Sailors from Sailors
where rating = 1 where rating = 2 where rating = 3

.....

select MIN(age) select MIN(age) select MIN(age)

from Sailors from Sailors from Sailors
where rating = 8 where rating = 9 where rating = 10

92
 Group-by and having clauses
Find the age of the youngest sailor for each rating level
Q44 select
from
rating, MIN(age)
Sailors
group by rating

Rule 1: a query that uses aggregate operators must use

only aggregate operators in the select clause, unless the
query contains a group by clause.

Rule 2: all non-aggregate columns in the target-list must

appear in the group by clause

93
 Group-by and having clauses
Find the age of the youngest sailor for each rating level

Q44 select
from
rating, MIN(age)
Sailors
group by rating

Find the age of the youngest sailor for each rating level,
such that there are at least two sailors with that rating

Q45 select
from
rating, MIN(age)
Sailors
group by rating
having count(*) > 1

94
 Group-by and having clauses

Rule 1: a query that uses aggregate operators must use only

aggregate operators in the select clause, unless the query
contains a group by clause.

Rule 2: all non-aggregate columns in the select clause must

appear in the group by clause

Rule 3: only non-aggregate columns used in the group by may

appear as non-aggregate columns in the having clause.

But we usually only use aggregate columns in the having clause anyway.

95
Aggregation: more examples
Find the number of reservations for each boat that was
reserved at least 3 times.

Q46*

Find the bid, name of each red boat that has been
reserved, along with an average rating and minimum age
of sailors who have reserved that boat.

Q47*

96
Aggregation: more examples
Find the number of reservations for each boat that was
reserved at least 3 times.
select bid, count(*) as num_reservations
Q46 from
group
Reserves
by bid
having count(*) >= 3

Find the bid, name of each red boat that has been
reserved, along with an average rating and minimum age
of sailors who have reserved that boat.
select B.bid, B.name, MIN(S.age), AVG(S.rating)
from Boats B, Reserves R, Sailors S
Q47 where B.bid = R.bid
and R.sid = S.sid
and B.color = 'red'
group by B.bid, B.name

97
Aggregation: more examples
Find the number of boat reservations by each sailor who is
at least 20 years old.

Q48*

Find the number of boat reservations by each sailor who

has made at least 2 reservations.
Q49*

98
Aggregation: more examples
Find the number of boat reservations by each sailor who is
at least 20 years old.
select R.sid, count (*)
Q48 from Reserves R, Sailors S
where R.sid = S.sid
and S.age >= 20
group by R.sid

Find the number of boat reservations by each sailor who

has made at least 2 reservations.
Q49 select sid, count(*)
from Reserves
group by sid
having count(*) >=2

99
Sorting results: more example
Retrieve all information about reservations, sorted by bid
select *
Q50 from Reserves
order by bid

Retrieve all information about reservations, sorted by bid

(descending), with ties broken by sid (ascending)
Q51 select *
from Reserves
order by bid desc, sid

Retrieve bids of all reserved boats, along with the number of

times they were reserved, and the average sailor rating, sorted
by average sailor rating
Q52
select R.bid, count(*) num_reservations, avg(S.rating) avg_rating
from Reserves R, Sailors S
where R.sid = S.sid
group by R.bid
order by avg_rating

100
Additional SQL features

• Outer joins

• Database modifications

101
 Running example
Animals (aid, name, hid) Habitats (hid, name)
aid name hid hid name
1 elephant 101 101 African savanna
2 giraffe 101
102 North Pacific ocean
3 whale 102
103 Eurasian tundra
4 dolphin 102
5 okapi null

List names of animals and names of their habitats

Q60 animal_name
elephant
habitat_name
African savanna
select A.name as animal_name,
H.name as habitat_name giraffe African savanna
from Animals A, Habitats H
where A.hid = H.hid
whale North Pacific ocean
dolphin North Pacific ocean

102
 Left outer join
Animals (aid, name, hid) Habitats (hid, name)
aid name hid hid name
1 elephant 101 101 African savanna
2 giraffe 101 102 North Pacific ocean
3 whale 102
103 Eurasian tundra
4 dolphin 102
5 okapi null
List names of animals and, if their habitat is known, also list
their habitat. animal_name habitat_name
elephant African savanna
Q61 giraffe African savanna
select A.name as animal_name,
whale North Pacific ocean
H.name as habitat_name
from Animals A left outer join Habitats H dolphin North Pacific ocean
on (A.hid = H.hid)
okapi null
103
 Right outer join
Animals (aid, name, hid) Habitats (hid, name)
aid name hid hid name
1 elephant 101 101 African savanna
2 giraffe 101
102 North Pacific ocean
3 whale 102
103 Eurasian tundra
4 dolphin 102
5 okapi null
List names of habitats and, if there are animals inhabiting
them, also list names of these animals.
animal_name habitat_name
elephant African savanna
Q62 giraffe African savanna
select A.name as animal_name,
H.name as habitat_name whale North Pacific ocean
from Animals A right outer join Habitats H
on (A.hid = H.hid) dolphin North Pacific ocean

null Eurasian tundra

104
 Full outer join
Animals (aid, name, hid) Habitats (hid, name)
aid name hid hid name
1 elephant 101 101 African savanna
2 giraffe 101
102 North Pacific ocean
3 whale 102
103 Eurasian tundra
4 dolphin 102
5 okapi null
List names of animals and names of habitats where they live. If no
habitat is known for an animal, or no animal is known to live in a
habitat - list these anyway. animal_name habitat_name
elephant African savanna
Q63 giraffe African savanna
select A.name as animal_name, whale North Pacific ocean
H.name as habitat_name
dolphin North Pacific ocean
from Animals A full outer join Habitats H
on (A.hid = H.hid) null Eurasian tundra
okapi null
105
 Join vs. outer join
Animals (aid, name, hid) Habitats (hid, name)
A.aid A.name A.hid H.hid H.name
1 elephant 101 101 African savanna
3 whale 102 102 North Pacific ocean
5 okapi null 103 Eurasian tundra

σ A.hid=H .hid (Animals × Habitats)

select *
from Animals A, Habitats H
where A.hid = H.hid

A.aid A.name A.hid H.hid H.name

1 elephant 101 101 African savanna
1 elephant 101 102 North Pacific ocean
1 elephant 101 103 Eurasian tundra
3 whale 102 101 African savanna
3 whale 102 102 North Pacific ocean
3 whale 102 103 Eurasian tundra

106
 Join vs. outer join
Animals (aid, name, hid) Habitats (hid, name)
A.aid A.name A.hid H.hid H.name
1 elephant 101 101 African savanna
3 whale 102 102 North Pacific ocean
5 okapi null 103 Eurasian tundra

select *
from Animals A full outer join Habitats H
on A.hid = H.hid

A.aid A.name A.hid H.hid H.name

1 elephant 101 101 African savanna
3 whale 102 102 North Pacific ocean join
left 5 okapi null null null
outer null null null 103 Eurasian tundra
right
outer

107
 Examples
Q64 select
from
A.name as animal_name, H.name as habitat_name
Animals A left outer join Habitats H
on (A.hid = H.hid)
where A.name like '%o%'

Q65 select H.name as habitat_name, count(*)

from Animals A full outer join Habitats H
on (A.hid = H.hid)
group by H.name
having count(*) = 1

compare results of Q65 and Q66!

Q66 select H.name as habitat_name, count(*)

from Animals A, Habitats H
where A.hid = H.hid
group by H.name
having count(*) = 1

108
 More examples
Animals (aid, name, hid) Habitats (hid, name)
aid name hid hid name
1 elephant 101 101 African savanna
2 giraffe 101
102 North Pacific ocean
3 whale 102
103 Eurasian tundra
4 dolphin 102
5 okapi null
List names of habitats that are not known to be inhabited
by any animals.
Q67

109
 More examples
Animals (aid, name, hid) Habitats (hid, name)
aid name hid hid name
1 elephant 101 101 African savanna
2 giraffe 101
102 North Pacific ocean
3 whale 102
103 Eurasian tundra
4 dolphin 102
5 okapi null
List names of habitats that are not know to be inhabited by
any animals.
Q67.a select
from
H.name as habitat_name
Habitats H left outer join Animals A
on (H.hid = A.hid)
where A.hid is null

Q67.b select
from
H.name as habitat_name
Animals A right outer join Habitats H
on (H.hid = A.hid)
where A.hid is null

110
 More examples
Animals (aid, name, hid) Habitats (hid, name)
aid name hid hid name
1 elephant 101 101 African savanna
2 giraffe 101
102 North Pacific ocean
3 whale 102
103 Eurasian tundra
4 dolphin 102
5 okapi null
List names of habitats that are not know to be inhabited
by any animals.
Q67.c select H.name as habitat_name
from Habitats H
select H.name as habitat_name
EXCEPT
where H.hid in ( select H.name
select hid from Habitats from Animals A, Habitats H
EXCEPT where A.hid = H.hid
select hid from Animals
)

111
 Database modifications
Insertion insert into R (A1, A2, ..., An) values (v1, v2, ..., vn)

Q68 insert into Animals (aid, name, hid)

values (6, 'zebra', 101)
Q69 insert into Animals (aid, name)
values (7, 'rhinoceros')

Deletion delete from R where <condition>

Q70 delete from Animals

where name = 'zebra'
Q71delete from Animals
where hid = 102
Q72 delete from Animals
where hid is null

Updates update R set <new value assignment> where <condition>

Q73update
set
Animals
hid = 101
Q74 update
set
Animals
name = upper(name),
where hid is null hid = 101

112