Fall 2020: CS 553 (Call no 39828) Distributed Computing Systems
Instructor:
Ajay Kshemkalyani
Email:
ajay@uic.edu
Class meeting times: TR 2:00-3:15pm (Tuesday and Thursday)
Room: on-line synchronous in Blackboard Collaborate Ultra
Office Hours in Blackboard Collaborate Ultra, immediately after class:
TR 3:15 - 3:30pm and beyond as needed
- Models: synchronous/asynchronous; shared memory/message-passing
- Global states and snapshots; time models and clock synchronization
- Distributed graph algorithms
- Group communication - total order and causal order
- Reasoning with knowledge
- Distributed shared memory - coherence, models, register constructions, atomic snapshots (applications to multicore architectures)
- Checkpointing, rollback recovery; distributed debugging
- Agreement and consensus (with malicious and non-malicious process behavior)
- Failure detectors
- Self-stabilizing systems
- Peer-to-peer systems, e.g., Chord, Tapestry, Content-Addressible Network, BitTorrent
- other current topics, e.g., sensor networks
See detailed table of contents of the textbook below by going to the link at Cambridge University Press or Amazon.
This course focuses on distributed algorithms and theoretical principles.
By the end of the course,
you will be able to appreciate that these algorithms have wide applications
in practical distributed systems.
- Textbook:
Distributed Computing: Principles, Algorithms, and Systems, by
Kshemkalyani and Singhal, Cambridge University Press, March 2011 edition
South Asian reprint edition, Dec 2010, ISBN-10: 1107648904, ISBN-13: 978-1107648906
- Course notes are here
- An Overview Chart
- Winners of the Dijkstra Award for Most Influential Paper in Distributed Computing, 2000-.
-
Selected papers and other material from the literature will be posted on the
web or distributed in class.
- Suggested Topics and Papers for class presentation (tentative list; to be updated)
- Self-Stabilization:
E.W. Dijkstra,
Self-stabilizing systems in spite of distributed control,
Communications of the ACM, vol. 17, no. 11, pp. 643-644, Nov. 1974.
plus Chapter 17 (Self-stabilization)
- Paxos and Raft:
L. Lamport,
Paxos Made Simple,
ACM SIGACT News (Distributed Computing Column) 32, 4 (121, December 2001)
51-58.
W. Bolosky et al, Paxos replicated state machines as the basis of a high-performance data store, NSDI 2011
D. Ongaro and J. Ousterhout, In Search of an Understandable Consensus Algorithm (Extended Version).
- C. Flanagan, S. Freund,
FastTrack: efficient and precise dynamic race detection, PLDI 2009.
-
W Cai, F He, X Lv, Y Cheng,
A transparent selective undo algorithm for collaborative editing,
IEEE 21st International Conference on
Computer Supported Cooperative Work in Design (CSCWD), 2017.
- Transactional Memory - software and hardware
- Conflict-free Replicated Data Types (CRDTs)
- Hadoop and Mapreduce:
J. Dean and S. Ghemawat,
MapReduce: Simplified Data Processing on Large Clusters,
OSDI'04: Sixth Symposium on Operating System Design and Implementation,
San Francisco, CA, December, 2004.
plus Hadoop
- Concurrent Data Structures
- Blockchain and bitcoin
The course format will be in two parts.
-
For the first part, the instructor will teach.
Attendance when the instructor is
teaching is not compulsory, but you must attend all the tests/exams.
However, if you miss class, it is your responsibility
to find out what was announced and what was covered, from other students.
-
The second part involves active student participation and is planned as follows.
Each presentation will be made by a team of 1 or 2 students,
depending on the final enrollment which will be known only in the third
week of class.
The class presentations will be on an assortment of topics of current interest.
Each group chooses a paper/topic from a list of topics provided around the
5th week of class.
This is only a starting point. Once you select a topic
from the list (to be provided), you may have to identify more basic or
fundamental papers on that specific topic for presentation.
Pick the most basic/ fundamantal papers that are
rich in new ideas.
They must also have algorithmic content.
Attendance when the student presentations are going on is compulsory.
There is also a programming project requirement (in a language of your choice).
Algorithm analysis and design (cs401) is suggested (but not required); or permission of the instructor.
The following is only a tentative breakup of the evaluation scheme
and will be finalized after the second week of class, depending on the final enrollment
in the course.
- midterm 1 (25%)
- midterm 2 (25%)
- Class presentation (15%)
- Term paper/programming project (35%)
The course grade is on the curve, i.e., this is relative grading - how you
perform with respect to the others in the class.
Tentative course progress chart (will be updated as we progress)
- Week 1: Introduction (Chapter 1)
- Week 2: Distributed graph algorithms (Chapter 5)
- Week 3: Chapter 5
- Week 4: Chapter 5, Chapter 2 (Models), Chapter 3 (Time); Project, Bloom Clock Intro slides, Bloom Clock paper
- Week 5: Chapter 3, Chapter 4 (Global state)
- Week 6: Chapter 6 (Message Ordering and Group Communication), Chapter 8 (Reasoning with Knowledge)
- Week 7: Chapter 8
Midterm 1 on Oct 8 (tentative):
Syllabus: Chapters 1, 5, 2, 3.1-3.5.1, 3.7, 3.9, 4-4.3, 6.1-6.1.3, 6.4-6.5.1, 6.6-6.7, 6.9-6.10
- Week 8: Chapter 14 (Consensus and Agreement)
- Week 9: Chapter 14, Chapter 12 (Distributed Shared Memory)
- Week 10:
- Oct 27: Midterm 2 (tentative):
Syllabus: Midterm 1 syllabus + 8.1-8.4, 8.6, 14.1-14.5.4, 14.6-14.6.4, 14.6.6, 12.1-12.2.7
- Oct 29: Self-Stabilization, Glenn
- Week 11:
- Nov 3: UIC holiday
- Nov 5: Static and Dynamic Race Detection, Naveen
- Week 12:
- Nov 10: Collaborative Group Editors, Saurabh, Yash
- Nov 12: Transactional Memory - software and hardware, Shreya
- Week 13:
- Nov 17: Conflict-free Replicated Data Types (CRDTs), Isacc
- Nov 19: Paxos and Raft, Aliza, Muhammad
- Week 14:
- Nov 24: Hadoop and Mapreduce, Prajwal, Rahul
- Week 15:
- Dec 1: Concurrent Data Structures, Vignan, Rohith . Programming project report due at 12 noon in Gradescope.
- Dec 3: Blockchain and bitcoin, Siddhanth