1

Lecture 1:
Introduction to databases
Timothy G. Griffin
Easter Term 2008 IB/Dip/IIG






www.cl.cam.ac.uk/Teaching/current/Databases/
2
Database Prehistory
Data entry
Storage and retrieval
Query processing
Sorting
3
Early Automation
Data management and application code
were all tangled together
Hard to modify
Hard to generalize
Many competing approaches
Data manipulation code written at very low
levels of abstraction
4
Our Hero --- E. F. Codd
Edgar F. "Ted" Codd ( August 23, 1923 - April 18, 2003) was a British computer scientist
who invented relational databases while working for IBM.
He was born in Portland, Dorset, studied maths
and chemistry at Oxford. He was a pilot in the
Royal Air Force during WWII. In 1948 he joined IBM
in New York as a mathematical programmer. He fled
the USA to Canada during the McCarthy period.
Later, he returned to the USA to earn a doctorate in CS
from the University of Michigan in Ann Arbor. He then
joined IBM research in San Jose.


His 1970 paper A Relational Model of Data for
Large Shared Data Banks changed everything.


In the mid 1990s he coined the term OLAP.
5
Database Management Systems (DBMSs)
Raw Resources (bare metal)
DBMS
Your Applications
Go Here
Database abstractions
allow this interface to
be cleanly defined and
this allows applications
and data management
systems to be
implemented
separately.
6
S
e
r
v
i
c
e

T
o
o
l
s

E
n
d

U
s
e
r
s

Service DB
Production DB
Development
DB
S
u
b
m
i
t
t
e
r
s

S
u
b
m
i
s
s
i
o
n

t
o
o
l
s

Add value
(computation)
Add value (review etc.)
Other archives
Q/C etc
Database
design
Releases
&
Updates
Releases
&
Updates
Today, Database Systems are Ubiquitous
Database system design from the European Bioinformatics Institute (Hinxton UK)
7
What is a database system?
A database is a large, integrated
collection of data
A database contains a model of
something!
A database management system
(DBMS) is a software system designed to
store, manage and facilitate access to the
database
8
What does a database system
do?
Manages Very Large Amounts of Data
Supports efficient access to Very Large
Amounts of Data
Supports concurrent access to Very
Large Amounts of Data
Supports secure, atomic access to Very
Large Amounts of Data
9
Databases are a Rich Area for
Computer Science
Programming languages and software
engineering (obviously)
Data structures and algorithms (obviously)
Logic, discrete maths, computation theory
Some of todays most beautiful theoretical results are
in finite model theory --- an area derived directly
from database theory
Systems problems: concurrency, operating
systems, file organisation, networks, distributed
systems
Many of the concepts covered in this course are classical --- they form
the heart of the subject. But the field of databases is still evolving and
producing new and interesting research (hinted at in lectures 11 & 12).
10
What this course is about
According to Ullman, there are three
aspects to studying databases:
1. Modelling and design of databases
2. Programming
3. DBMS implementation
This course addresses 1 and 2

11
Course Outline
1. Introduction
2. Entity-Relationship Model
3. The Relational Model
4. The Relational Algebra
5. The Relational Calculus
6. Schema refinement: Functional
dependencies
7. Schema refinement: Normalisation
8. Transactions
9. Online Analytical Processing (OLAP)
10. More OLAP
11. Basic SQL and Integrity Constraints
12. Further relational algebra, further SQL




12
Recommended Reading
Date, An introduction to database systems, 8
th
ed.
Elmasri & Navathe, Fundamentals of database systems, 4
th
ed.
Silberschatz, Korth & Sudarshan, Database system concepts, 4
th

ed.
Ullman & Widom, A first course in database systems.
OLAP
DB2/400: Mastering Data Warehousing Functions. (IBM Redbook) Chapters 1 &
2 only. http://www.redbooks.ibm.com/abstracts/sg245184.html

Data Warehousing and OLAP
Hector Garcia-Molina (Stanford University)
http://www.cs.uh.edu/~ceick/6340/dw-olap.ppt

Data Warehousing and OLAP Technology for Data Mining Department of
Computing
London Metropolitan University
http://learning.unl.ac.uk/csp002n/CSP002N_wk2.ppt


13
Some systems to play with
1. mysql:
www.mysql.org
Open source, quite powerful
2. PostgreSQL:
www.postgresql.org
Open source, powerful
3. Microsoft Access:
Simple system, lots of nice GUI wrappers
4. Commercial systems:
Oracle 10g (www.oracle.com)
SQL Server 2000 (www.microsoft.com/sql)
DB2 (www.ibm.com/db2)
14
Database system architecture
It is common to describe databases in two ways
The logical level:
What users see, the program or query language
interface,
The physical level:
How files are organised, what indexing mechanisms
are used,
It is traditional to split the logical level into two:
overall database design (conceptual) and the
views that various users get to see
A schema is a description of a database
15
Three-level architecture
Conceptual
Schema

Physical
level
Conceptual
level
Internal
Schema
External
Schema 1
External
Schema 2
External
Schema n
External
level
16
Logical and physical data
independence
Data independence is the ability to change the
schema at one level of the database system
without changing the schema at the next higher
level
Logical data independence is the capacity to
change the conceptual schema without
changing the user views
Physical data independence is the capacity to
change the internal schema without having to
change the conceptual schema or user views
17
Database design process
Requirements analysis
User needs; what must database do?
Conceptual design
High-level description; often using E/R model
Logical design
Translate E/R model into (typically) relational schema
Schema refinement
Check schema for redundancies and anomalies
Physical design/tuning
Consider typical workloads, and further optimise
Next Lecture
18
The Fundamental Tradeoff of Database
Performance Tuning
De-normalized data can often result in faster
query response
Normalized data leads to better transaction
throughput, and avoids update anomalies
(corruption of data integrity)

What is more important in your database --- query response
or transaction throughput? The answer will vary.
What do the extreme ends of the spectrum look like?
Yes, indexing data can speed up transactions, but this just proves
the point --- an index IS redundant data. General rule of thumb:
indexing will slow down transactions!
19
A Theme of this Course:
OLTP vs. OLAP
OLTP = Online Transaction Processing
Need to support many concurrent transactions
(updates and queries)
Normally associated with the operational database
that supports day-to-day activities of an organization.
OLAP = Online Analytic Processing
Often based on data extracted from operational
database, as well as other sources
Used in long-term analysis, business trends.

20
S
e
r
v
i
c
e

T
o
o
l
s

E
n
d

U
s
e
r
s

Service DB
Production DB
Development
DB
S
u
b
m
i
t
t
e
r
s

S
u
b
m
i
s
s
i
o
n

t
o
o
l
s

Add value
(computation)
Add value (review etc.)
Other archives
Q/C etc
Database
design
Releases
&
Updates
Releases
&
Updates
Design Heterogeneity
Database system design from
the European
Bioinformatics Institute (Hinxton UK)
De-normalized Derived Tables
--- for fast access
Normalized Tables