1/24/25

CS3700
Introduction to Database Systems
Perspective - Big Picture
Prof P Sreenivasa Kumar
Department of CSE, I I T Madras

Introduction to Database Systems

CS3700
Objectives:
Introduce the learner to the fundamental ideas /
principles of relational databases
and relational database management systems.

Next few slides:

A perspective of the “data” space

Prof P Sreenivasa Kumar, 2

Department of CS&E, IITM.

Kinds of Data
• Various kinds of Data we encounter everyday
– Enterprise (business) data
– Structured data
– Documents / webpages – Text data
– Objects – JSON; XML – (label, value) pairs data
– Semi-structured data
– Voice / music / image / video data
– Unstructured data
• This course focuses on structured data

Prof P Sreenivasa Kumar, 3

Department of CS&E, IITM.

1
 1/24/25

Data at Large Enterprises

• Large Enterprises (Businesses)
– Need to deal with multiple types of data
– E-commerce company
– Product specs, images, demo videos, customer profiles …
– Transactions data

– Multiples sources of data

– Each division – its own data
– External data – stock prices, media reports etc
– Data Integration – a necessity

• Need multiple DBs and a Data Warehouse

Prof P Sreenivasa Kumar, 4
Department of CS&E, IITM.

Data Warehouses

Prof P Sreenivasa Kumar, 5

Department of CS&E, IITM.

Data Mining vs Querying

• Data Mining – Knowledge Discovery in Databases
– Discover hither-to unknown ‘knowledge’ in data
– Knowledge – associations / clusters / classifications ...
– Custom-designed algorithms

• DB Querying
– Ask for details present in data or can be derived
– Use logic-based expression languages – SQL/OQL etc
– Ask for aggregated information
– How many ‘fiction’ books got issued to ‘CS’ students in the
last quarter ?
Prof P Sreenivasa Kumar, 6
Department of CS&E, IITM.

2
 1/24/25

OLAP - OLTP
• OLAP (Online analytical processing) queries:
– Ask for aggregated information from data
– Needs summarization of large amount of data
– How many ‘cosmetics’ products were sold in all the stores located
in the ‘southern’ region of the country in the last 2 quarters ( Big
Basket or Amazon )
– Specialized architecting of data is needed
– Done on warehouse data using “Data Cubes”

• OLTP (Online transaction processing)

– Process a transaction (sale / credit card use / reservation)
quickly – concurrency needs attention
– An important functionality of DB servers

Prof P Sreenivasa Kumar, 7

Department of CS&E, IITM.

Text data - IR Systems

• IR - Information Retrieval Systems
– Typified by Search Engines – Web/Enterprise
– Query – list of keywords – (avg : 2.4 words/query)
– Intention is imprecise
– Relational queries are very precise ( to see later)
– Results – a ranked list(s) of documents
– Ordered by relevance to the Query
– Hugely successful
– Modern IR systems – not just a repository of docs!
– Needs a separate course
Prof P Sreenivasa Kumar, 8
Department of CS&E, IITM.

OO Languages and Databases

• Object-Oriented (OO) programming
– Deeply structured objects – nesting is common
– Rectangle – set of 4 point objects; Point – a pair
– Objects are not ‘persistent’ !
• Data persistence is one of the strength of DBs
• Two approaches
– Take persistence to OO Languages
– Object Database Systems – OQL
– Take ‘rich structures’ to databases
– Object-Relational Databases (ORDBMSs)
Prof P Sreenivasa Kumar, 9
Department of CS&E, IITM.

3
 1/24/25

Knowledge Bases
• Knowledge Bases
– Founded on mathematical logic based languages
– Inferences / reasoning incorporated
– Every surgeon is a doctor; John is a surgeon and hence a doctor
– Data : modeled as collection of ‘statements’
– Metadata: Terms/Vocabulary of the domain
– Names of entity types of interest, properties of interest
for these entities; binary relationships among entities
– Modeled as Ontologies to capture domain knowledge
– Can be viewed as graphs – ‘Knowledge Graphs’
– More in the TAO course….
Prof P Sreenivasa Kumar, 10
Department of CS&E, IITM.

10

NoSQL Databases
• Non-relational databases / Not only SQL
– Offer flexibility – no rigid row/col structure
– Semi-structured + other simpler data models
– Need: faster response to certain queries, quick data
model updates, quick app development, horizontal
scale-out ..
– System ‘availability’ over ‘consistency’ of reads
– (-ve) No standard query language to access data
– Multiple types of DBs
– JSON / XML, key-value, columnar, graph etc

Prof P Sreenivasa Kumar, 11

Department of CS&E, IITM.

11

Types of NoSQL DBs

• JSON / XML DBs
– aka ‘Document-oriented’ / ‘document’ DBs
– Semi-structured / non-homogenous data model
• Key-value Stores
– Value – single or a collection of values (opaque)
– Like hash-tables – give key, get value – simple
• Graph DBs – Nodes and labeled edges
– Query language support – SPARQL etc
• Column-oriented DBs – fast access of col data
Prof P Sreenivasa Kumar, 12
Department of CS&E, IITM.

12

4
 1/24/25

Course Structure
Course has 8 modules (Total #slides 451)
Module Name #Slides

Introduction 24

Entity/Relationship Models 38

Relational Model 64

Tuple Relational Calculus 20

SQL Standard 66

Indexes and File Storage 68

Normal Forms 64

Algorithms for Rel Ops 33

Transaction Processing 74

Prof P Sreenivasa Kumar, 13

Department of CS&E, IITM.

13

Course Organization
• Problem Sessions – 5 or 6 ( tutorials )
– Problem sheet – start of the class or a day before
– Students to solve – discussion and questions – OK
– No marks – no judging; answers posted later
– Course TAs would be present - aid discussions
• Design and Dev assignment – group/individual
– Running assignment – with 3 or 4 stages
• Exams
– Mid-sem Exam / Two Quizzes – TBD
– End-sem Exam as per schedule.
Prof P Sreenivasa Kumar, 14
Department of CS&E, IITM.

14