CS 382: Network-Centric Computing

Course Overview

Dr. Zafar Ayyub Qazi

Spring 2021

1
 Agenda for Today’s Class

1. What is this course about?

2. What is the course structure, grading, workload?

3. Who are the course staff?

4. What do I expect from you?

2
Send/Receive data over a network

3
 How are these apps architected?
Internet is a key enabler for these applications

User Device Remote services

Internet

4
 What is the Internet?
l Take 2-3 mins and write down what you think is
the Internet

l Write your responses under the Chat section

5
 What is the Internet?
l The Internet is not a particular kind of network
l There are many different types of networks
l Cellular, WiFi, Ethernet, Optical networks, etc.

l The Internet ties different networks together

l The Internet

6
What does the Internet look like?

• Interconnects different networks

(>18,000)

• > 4.6 billion Internet users (~60%

of world’s population)

• > 30 billion devices (PCs,

laptops, smart phones, IoTs, etc)

7
 First Half of the Course
l Study fundamental network communication
concepts via the case study of the Internet

8
 First Half: Fundamental Questions
l How to write applications that can communicate over the
network?
l Socket programming, Web and HTTP Assignment 1
Part a

l How to reliably transfer data over an unreliable network?

l Design a reliable transport Assignment 1
Part b

l How to control congestion on the Internet?

l How to route packets over Internet? Assignment 2

l Routing approaches, addressing, forwarding

9
 End-End story

User Remote
device services

Internet

10
If you are designing a Facebook-like service,
what would be your design goals?

11
 Service Provider Design Goals
l Scalability
l Scale to support millions/billions of users

l Availability
l Services should be highly available, despite failures

l Low latency
l Customers should be able to access services with low
latency

12
 To scale, applications are hosted
in large data centers
l 10,000+ servers common in large-scale data centers
l Single “applications” spread across many servers

13
Datacenters with 100,000+ Servers

14
 Distributed Applications
l Typical applications have many distributed
components
l E.g., a bing search query touches > 100 machines

l Typical applications access large amounts of data

which is distributed across many storage servers
l E.g., a Facebook page request can result in > 1700
distinct data items being fetched from different servers
 Latency Matters

Impacts user experience & operator revenue

 Failure are common in large systems
Failure statistics from clusters in Google data center*

*Design, Lessons and Advice for Building Large Scale Distributed Systems by Jeff Dean, Google Senior Fellow and lead of Google AI, 2009

17
Second Half: Fundamental Questions
l How to coordinate between different components of a
distributed application?
l Time synchronization, logical clocks, election algorithms

l How to distribute data across application components?

l DHTs, consistent hashing, lookups
Assignment 3

l How to design for fault tolerance?

l Replication, consistency and failure handling

Assignment 3

18
 Course Organization
Make sure you carefully go over the course outline

You are responsible for reading everything in the course outline

19
 Course Teaching Methodology
l Synchronous (live lectures) on Zoom
l Lectures will be recorded
l Recordings will be uploaded on YouTube

l Piazza for course-related discussion

l Announcements, questions, clarificatons
l Please signup if you haven’t already done so

20
 Learning the Material
l Attend lectures
l Participate in class discussions
l You are responsible for everything covered in class

l Read the book

l Computer Networking: A Top-Down Approach, 6th Edition, 2016, by
J. Kurose and K. Ross
l Distributed Systems: Principles and Paradigms, 3rd Edition, by
Andrew S. Tanenbaum and Maarten Van Steen.
l Other optional books recommended on the course outline
l Books available on LMS under Books folder
l Recommended readings for each lecture given in course outline
l You will not be tested on material we didn't cover in lectures
21
 Learning the Material
l Post questions on Piazza
l Help each other by answering queries
l Don’t send course staff direct emails

l Attend tutorials

22
 Course Grading
l Programming assignments
l Total 35% of the grade
l Three programming assignments
l Programming language: Python

l Exams:
l Final: 30% (Comprehensive)

l Quizzes
l 35% of the grade, N-2 policy, Announced in outline

23
 Programming Assignments (PA)
l Goal: Practice designing, implementing and
debugging networked applications and protocols

l PA1: Chat application that communicate reliably

l PA2: Implementing routing protocols

l PA3: A fault-tolerant peer-peer file sharing app.

24
 Policies: Write your own code
l All assigned work must be done individually (unless
specified otherwise)

l Students must be prepared to explain any program code

they submit

l All submissions are subject to plagiarism detection

l Students cannot copy code from the Internet

l Students are strongly advised that any act of plagiarism

will be reported to the Disciplinary Committee
25
 Policies: Write your own code
l All assigned work must be done individually (unless
specified otherwise)

e !
l
iz
Students must be prepared to explain any program code
r
they submit
g ia
All submissions are subjectla
l

’t P to plagiarism detection

o
Students cannot n
copy code from the Internet
D
l

l Students are strongly advised that any act of plagiarism

will be reported to the Disciplinary Committee
26
 Integrity Violations
l MOSS: plagarism detection tool for programming
assignments

l Last year, more than 50 students were caught for

plagarism in programming assignments of CS 382

l Cases will be reported to Disciplinary commiteee

l Penalty could arrange from 0 in the assignment to

an F grade, and in the worst case explusion
27
 Late Day Policy for PA
l 90% for work submitted up to 24 hours late
l 80% for work submitted up to 2 days late
l 70% for work submitted up to 3 days late
l 60% for work submitted up to 5 days late
l 50% for work submitted after 5 days late

l In addition, you have five “free” late days during

the semester

28
 Warnings
l Assignments will take a lot of time
l Need to understand the problem and protocol, and then
carefully design
l Can take 5x more time to debug than “initially program”
l Start early

l Don’t expect to get through the course by studying

just a few days before the exams

29
Course Staff

30
 Instructor
l Instructor
l Dr. Zafar Ayyub Qazi
l Assistant Professor at LUMS, 2018-present
l Email: zafar.qazi@lums.edu.pk
l Webpage: http://web.lums.edu.pk/~zafar/
l Academic background
l Postdoctoral Scholar at UC Berkeley, 2016-2017
l PhD in CS, SUNY Stony Brook, 2010-2015
l BSc in CS, LUMS, 2009
l Area of Research
l Distributed Systems and Networks
31
 Teaching Assistants
l Maleeha Masood
l Muhammad Basit Awan
l Hayat Awais Malik
l Taimoor Tariq
l Talal Touseef
l Syeda Mashal Abbas Zaidi
l Adnan Abbas
l Sadia Zubair
l Abdul Monum

32
What do I expect from you?

33
 I Expect Three Things
l Try to attend every class

l Ask questions when you don’t understand

l Or when you want to understand better

l Answer (or think about) questions when I ask them

l Even if you aren’t sure of the answer

34
Questions?

35
 Next Class
l Overview of the Internet

36
Thank you!

37