Motivation Outline

Course Overview and Administrivia


A Primer on Databases
On the Verge of A Disruptive Century:
Breakthroughs

Gene
Ubiquitous
Sequencing and
Computing
Biotechnology

Smaller, Faster,
Cheaper Sensors

Faster
Communication
 The amount of data is only growing…
A Common
1.2 Zettabytes (1ZB = 1021 BTheme isTB)
or 1 Billion Data
in 2010
 We Live in a World of Data
 Nearly 500 Exabytes per day are generated by the Large Hadron
Collider experiments (not all recorded!)

 2.9 million emails are sent every second

 20 hours of video are uploaded to YouTube every minute

 24 PBs of data are processed by Google every day

 50 million tweets are generated per day

 700 billion total minutes are spent on Facebook each month

 72.9 items are ordered on Amazon every second

 Data and Big Data
 The value of data as an organizational asset is widely recognized

 Data is literally exploding and is occurring along three main dimensions

• “Volume” or the amount of data
• “Velocity” or the speed of data
• “Variety” or the range of data types and sources

 What is Big Data?

 It is the proliferation of data that floods organizations on a daily basis

 It is high volume, high velocity, and/or high variety information assets

 It requires new forms of processing to enable fast mining, enhanced decision-

making, insight discovery and process optimization
 Share
Store

What Do We Do With Data and Big Data?

Mine
Query

more!
Encrypt
…. and

We want to do these seamlessly and fast...

Using Diverse Interfaces & Devices

Mobile Devices
Computers

…and even appliances

Consumer Electronics Personal Monitors and

Sensors

We also want to access, share and process our data from all of our devices,
anytime, anywhere!
Data is Becoming Critical to Our Lives

Health Science
Domains
Education of Data
Work

Environment Finance

… and more
 Why Studying Databases?
 Data is everywhere and is critical to our lives

 Data need to be recorded, maintained, accessed and manipulated

correctly, securely, efficiently and effectively
 At the “low end”: scramble to web-scale (a mess!)
 At the “high end”: scientific applications

 Database management systems (DBMSs) are indispensable software

for achieving such goals

 The principles and practices of DBMSs are now an integral part of

computer science curricula
 They encompass OS, languages, theory, AI, multimedia, and logic,
among
As such, others
the study of database systems can prove to be richly rewarding in more
ways than one!
Motivation Outline
Course Overview and Administrivia

A Primer on Databases

 Course Objectives
In this course we aim at studying:

Big Data,
Hadoop,
How to construct BigTable, parallel
buffer and disk and distributed
How to refine space managers, DBMSs, NoSQL
and speed up query optimizers, and NewSQL
How to query data retrieval and concurrency databases
and manipulate and and crash
How to design databases manipulation recovery
and implement managers for
databases from
DBMSs
‘cradle-to-grave’

Application-Centric Systems-Centric & Theory-Centric Advanced Topics

(A Brief Overview)
 List of Topics
Considered: a reasonably critical and
comprehensive understanding.
.1.
Thoughtful: fluent, flexible and efficient The Entity-Relationship Model
understanding.
.2.
Masterful: a powerful and illuminating The Relational Model
understanding.
.3.
Relational Algebra and Calculus
.4.
SQL
.5.
Data Storage and Organization
.6.
Tree-Based and Hash-Based Indexing
.7.
Query Evaluation and Optimization
.9.
Concurrency Control and Crash Recovery
.10.
Advanced Topics: Distributed Databases,
Hadoop, and NoSQL and NewSQL
Databases
 Learning Outcomes
 After finishing this course you will be able to:

1. Describe a wide range of data involved in real-world organizations using the entity-
relationship (ER) data model

2. Explain how to translate an ER diagram into a relational database

3. Analyze and apply a formal query language, relational calculus and algebra

4. Indicate how SQL builds upon relational calculus and algebra and effectively apply
SQL to create, query and manipulate relational databases
5. Design and develop multi-tiered, full-fledged standalone and web-based
applications with back-end databases
6. Appreciate how DBMSs create, manipulate and manage files of fixed-length and
variable-length records on disks
 Learning Outcomes
 After finishing this course you will be able to:

7. Create and operate various static and dynamic tree-based (e.g., ISAM and B+ trees)
and hash-based (e.g., extendable and linear hashing) indexing schemes
8. Explain and evaluate various algorithms for relational operations (e.g., join) using
techniques such as iteration, indexing and partitioning
9. Analyze and apply different query evaluation plans and describe the various tasks of
a typical relational query optimizer
10. Describe how transactions can be interleaved correctly, and indicate how a DBMS
can ensure atomicity and durability when systems fail or entirely crash
11. Identify alternative architectures for distributed databases, and describe how data
can be partitioned and distributed across networked nodes of a DBMS
12. Appreciate the scale of Big Data, discuss some popular analytics engines for Big
Data processing and denote the applicability of NoSQL databases for Big Data storage
 Teaching Methods, Assignments
and Projects
26 Lectures
• Motivate learning
• Provide a framework or roadmap to organize the information of the course
• Explain subjects and reinforce the critical big ideas
14 Recitations
• Get you to reveal what you do not understand, so we can help you
• Allow you to practice skills you will need to become an expert

5 Assignments
• We will have 5 assignments which involve problem solving and span most of the
topics that we discuss in the class
3 Projects
• We will have 3 projects which involve using Postgres, SQL, Python, and Django
 Some Rules on the Projects
 For all the projects (except the final one), the following rules
apply:
 If you submit one day late, 25% will be deducted from your
project score

 If you are two days late, 50% will be deducted

 The project will not be graded (and you will receive a zero
score) if you submit more than two days late

 There will be a 3-grace-day quota

 Assessment Methods

How do we measure learning?

Type # Weight
Projects 3 40%
Exams 2 30%
Problem Solving Assignments 5 15%
Quizzes 2 10%
Class/Recitation Participation and 42 5%
Attendance
 Target Audience, Prerequisites
and Textbook
 Target Audience:
 Juniors and Seniors

 Prerequisites:
 15-121 and 15-213
 Students should have a basic knowledge of data
structures, algorithms, computer systems and
programming languages like C, C++ and Python

 Textbook:
 Raghu Ramakrishnan and Johannes Gehrke, "Database
Management Systems", Third Edition, McGraw-Hill, 2002
Motivation Outline
Course Overview and Administrivia

A Primer on Databases


 A Motivating Scenario
 Qatar Foundation (QF) has a “large” collection of data (say 500GB) on
employees, students, universities, research centers, etc.,

 This data is accessedPerformance

concurrently(Concurrency Control)
by several people

Performance (Response Time)

 Queries on data must be answered quickly

Correctness (Consistency)
 Changes made to the data by different users must be applied consistently

Correctness (Security)
 Access to certain parts of data (e.g., salaries) must be restricted
Correctness (Durability and Atomicity)
 This data should survive system crashes/failures
 Managing Data using File Systems
 What about managing QF data using local file systems?
 Files of fixed-length and variable-length records as well as formats
 Main memory vs. disk
 Computer systems with 32-bit addressing vs. 64-bit addressing schemes
 Special programs (e.g., C++ and Python programs) for answering user questions
 Special measures to maintain atomicity
 Special measures to maintain consistency of data
 Special measures to maintain data isolation
 Special measures to offer software and hardware fault-tolerance
 Special measures to enforce security policies in which different users are
granted different permissions to access diverse subsets of data

This becomes tedious and inconvenient, especially at large-scale, with

evolving/new user queries and higher probability of failures!
 Data Base Management Systems
 A special software is accordingly needed to make the preceding
tasks easier

 This software is known as Data Base Management System (DBMS)

 DBMSs provide automatic:

 Data independence
 Efficient data access
 Data integrity and security
 Data administration
 Concurrent access and crash recovery
 Reduced application development and tuning time
 Some Definitions
 A database is a collection of data which describes one or many
real-world enterprises
 E.g., a university database might contain information about entities like
students and courses, and relationships like a student
enrollment in a course

 A DBMS is a software package designed to store and

manage databases
 E.g., DB2, Oracle, MS SQL Server, MySQL and Postgres

 A database system = (Big) Data + DBMS + Application Programs

 Data Models
 The user of a DBMS is ultimately concerned with some real-world
enterprises (e.g., a University)

 The data to be stored and managed by a DBMS describes various

aspects of the enterprises
 E.g., The data in a university database describes students, faculty and
courses entities and the relationships among them

 A data model is a collection of high-level data description constructs

that hide many low-level storage details

 A widely used data model called the entity-relationship (ER) model

allows users to pictorially denote entities and the relationships
among them
 The Relational Model
 The relational model of data is one of the most widely used
models today

 The central data description construct in the relational model

is the relation

 A relation is basically a table (or a set) with rows (or records or

tuples) and columns (or fields or attributes)

 Every relation has a schema, which describes the columns

of a relation

 Conditions that records in a relation must satisfy can be specified

 These are referred to as integrity constraints
 The Relational Model: An Example
 Let us consider the student entity in a university database
Students Schema

Students(sid: string, name: string, login: string, dob: string, gpa: real)

An attribute, field or column

Integrity Constraint: Every student has a unique sid value

sid name login dob gpa

A record, tuple 512412 Khaled [email protected] 18-9-1995 3.5
or row
512311 Jones [email protected] 1-12-1994 3.2

512111 Maria [email protected] 3-8-1995 3.85

An instance of a Students relation

 Levels of Abstraction
 The data in a DBMS is described at three levels of abstraction,
the conceptual (or logical), physical and external schemas
View 1 View 2 View 3
 The conceptual schema describes
data in terms of a specific data model Conceptual Schema
(e.g., the relational model of data)
Physical Schema
 The physical schema specifies how data
described in the conceptual schema are
stored on secondary storage devices Disk

 The external schema (or views) allow data

access to be customized at the level of individual users or group of
users (views can be 1 or many)
 Views
 A view is conceptually a relation

 Records in a view are computed as needed and usually not

stored in a DBMS

 Example: University Database

Conceptual Schema Physical Schema External Schema (View)
• Students(sid: string, name: • Relations stored as heap files Students can be allowed to find
string, login: string, dob: string, • Index on first column of out course enrollments:
gpa:real) Students • Course_info(cid: string,
• Courses(cid: string, enrollment: integer)
cname:string, credits:integer)
• Enrolled(sid:string, cid:string, Can be computed from the relations in
grade:string) the conceptual schema (so as to avoid
data redundancy and inconsistency).
 Iterating: Data Independence
 One of the most important benefits of using a DBMS is
data independence

 With data independence, application programs are insulated

from how data are structured and stored

 Data independence entails two properties:

 Logical data independence: users are shielded from changes in the
conceptual schema (e.g., add/drop a column in a table)
 Physical data independence: users are shielded from changes in the
physical schema (e.g., add index or change record order)
 Queries in a DBMS
 The ease with which information can be queried from a database
determines its value to users

 A DBMS provides a specialized language, called the query language,

in which queries can be posed

 The relational model supports powerful query languages

 Relational calculus: a formal language based on mathematical logic
 Relational algebra: a formal language based on a collection of
operators (e.g., selection and projection) for manipulating relations
 Structured Query Language (SQL):
 Builds upon relational calculus and algebra
 Allows creating, manipulating and querying relational databases
 Can be embedded within a host language (e.g., Java)
 Concurrent Execution and Transactions
 An important task of a DBMS is to schedule concurrent accesses to
data so as to improve performance
T1 T2
R(A)
W(A) An atomic sequence of database actions (read/writes)
R(B)
W(B)
is referred to as “transaction”
R(C)
W(C)

 When several users access a database concurrently, the DBMS must

order their requests carefully to avoid conflicts
 E.g., A check might be cleared while account balance is being computed!

 DBMS ensures that conflicts do not arise via using a locking protocol
 Shared vs. Exclusive locks
 Ensuring Atomicity
 Transactions can be interrupted before running to completion for a
variety of reasons (e.g., due to a system crash)

 DBMS ensures atomicity (all-or-nothing property) even if a crash

occurs in the middle of a transaction

 This is achieved via maintaining a log (i.e., history) of all writes

to the database
 Before a change is made to the database, the corresponding log entry is
forced to a safe location (this protocol is called Write-Ahead Log or WAL)
 After a crash, the effects of partially executed transactions are undone
using the log
The Architecture of a Relational DBMS
Web Forms Application Front Ends SQL Interface

SQL Commands

Plan Executer Parser Query

Evaluation
Operator Evaluator Optimizer Engine

Transaction Files and Access Methods

Manager Recovery
Buffer Manager
Lock Manager
Manager Disk Space Manager
Concurrency Control DBMS

Database Index Files System Catalog

Data Files
 People Who Work With Databases
 There are five classes of people associated with databases:
1. End users
 Store and use data in DBMSs
 Usually not computer professionals
2. Application programmers
 Develop applications that facilitate the usage of DBMSs for end-users
 Computer professionals who know how to leverage host languages, query
languages and DBMSs altogether
3. Database Administrators (DBAs)
 Design the conceptual and physical schemas
 Ensure security and authorization
 Ensure data availability and recovery from failures
 Perform database tuning
4. Implementers
 Build DBMS software for vendors like IBM and Oracle
 Computer professionals who know how to build DBMS internals
5. Researchers
The Architecture of a Relational DBMS
Web Forms Application Front Ends SQLApplication
Interface
End Users (e.g., university staff, travel agents, etc.)
Programmers & DBAs
SQL Commands

Plan Executer Parser Query

Evaluation
Operator Evaluator Optimizer Engine

Transaction Files and Access Methods

Manager Recovery
Implementers and Researchers
Buffer Manager
Lock Manager
Manager Disk Space Manager
Concurrency Control DBMS

Database Index Files System Catalog

Data Files
 Summary
 We live in a world of data

 The explosion of data is occurring along the 3Vs dimensions

 DBMSs are needed for ensuring logical and physical data

independence and ACID properties, among others

 The data in a DBMS is described at three levels of abstraction

 A DBMS typically has a layered architecture

 Summary
 Studying DBMSs is one of the broadest and most exciting
areas in computer science!

 This course provides an in-depth treatment of DBMSs with an

emphasis on how to design, create, refine, use and build
DBMSs and real-world enterprise databases

 Various classes of people who work with databases hold

responsible jobs and are well-paid!