Semester: Fall 2012
Lecture Time: Tuesday/Thursday, 11:25AM - 12:40PM
Lecture Location: Stuart Building 239
Professor: Dr. Ioan Raicu (firstname.lastname@example.org, 1-312-567-5704)
Office Hours Time: Thursday, 12:45PM-1:45PM
Office Hours Location: Stuart Building 237D
Teaching Assistant: Tonglin Li (email@example.com)
Office Hours Time: Monday/Tuesday, 1PM-2PM
Office Hours Location: Stuart Building 006
The specific topics that this course will cover are: scheduling in multiprocessors, memory hierarchies, synchronization, concurrency control, fault tolerance, data parallel programming models, scalability studies, distributed memory message passing systems, shared memory programming models, tasks, dependence graphs and program transformations, parallel I/O, applications, tools (Cuda, Swift, Globus, Condor, Amazon AWS, OpenStack, Cilk, gdb, threads, MPICH, OpenMP, Hadoop, FUSE), SIMD, MIMD, fundamental parallel algorithms, parallel programming exercises, parallel algorithm design techniques, interconnection topologies, heterogeneity, load balancing, memory consistency model, asynchronous computation, partitioning, determinacy, Amdahl's Law, scalability and performance studies, vectorization and parallelization, parallel programming languages, and power. Some of these topics are covered in more depth in the graduate courses focusing on specific sub-domains of distributed systems, such as Advanced Operating Systems (CS550), Parallel Computing (CS546), Cloud Computing (CS553), Data-Intensive Computing (CS554), Advanced Computer Architecture (CS570), and Fault Tolerant Computing (CS595).
While this CS495 course is not a pre-requisite to any of the graduate level courses in distributed systems, both undergraduate and graduate students who wish to be better prepared for these courses could take this CS495 course. Undergraduate students are highly encouraged to take CS495 prior to any of the graduate level courses in distributed systems. Graduate students who have already taken CS546, CS550, CS553, CS554, CS570, or CS595 should not take this CS495 class. Furthermore, this CS495 class should not be taken concurently with CS546, CS550, CS553, CS554, CS570, or CS595.
Many of these graduate courses are part of the Master of Computer Science Specialization in Distributed and Cloud Computing. This CS495 course is also a part of the Undergraduate Specialization in Data Science and the Specialization in Distributed and Cloud computing.
The course syllabus can be found here.
Unique Opportunity -- Hands-on Practical Experience
An important component of learning is to gain hands-on experience that a textbook just cannot teach. A portion of this course will cover practical aspects of distributed systems. For example, enrolled students will participate in the design, assembling, configuring, and benchmarking of a real cluster. The students would be exposed to practical issues in real cluster design, such as hardware tradeoffs, different operating systems, local and distributed storage, networking, virtualization, and grid/cloud middleware. The students will work in teams to build workstations/servers from scratch. The software stack will include Linux, XEN, Globus, Condor, OpenStack, NFS, PVFS, MPI, Swift, and Hadoop. This new cluster will then be used in subsequent assignments.
Another set of assignments will deal with real cloud systems, such as Google App Engine, Amazon EC2/S3, and Hadoop (MapReduce framework).
I am also assembling a team of undergraduate students to compete in the Supercompiting 2013 Student Cluster Competition. The team has three slots (of the total 6 slots) available.
Finally, I am also looking for an undergraduate student to join my DataSys Laboratory for a paid assistantship. If you are thinking about graduate school, or are excited about the opportunity to work at some of the largest technology companies (e.g. Microsoft, Google, Amazon, Facebook, Twitter, etc), then working in the DataSys Lab for several semesters will give you a significant advantage! Feel free to contact other students in my lab for feedback about the kinds of projects they are working on.
|08-23-2012||Introduction to Distributed Systems||DCC Ch. 1|
|08-28-2012||Introduction to Distributed Systems||Project #1 (Build Cluster)|
|08-30-2012||Distributed System Models and Enabling Technologies|
|09-04-2012||Distributed System Models and Enabling Technologies|
|09-06-2012||Distributed System Models and Enabling Technologies||Project #1 due
Project #2 (Benchmarking)
|09-11-2012||Distributed System Architectures||DSPD Ch. 2|
|09-13-2012||Distributed System Architectures|
|09-18-2012||Parallel Programming Systems and Models
|09-20-2012||Parallel Programming Systems and Models|
|09-25-2012||Parallel Programming Systems and Models||Project #2 due
Project #3 (MapReduce)
|09-27-2012||MapReduce||DCC Ch. 6
MapReduce vs DBMS
|10-04-2012||Workflow Systems||DCC Ch. 6
Swift Workflow System
|Exam #1||DCC 1, 6, DSPD 2 + External Reading ((MapReduce and Workflows)||Exam #1|
|10-11-2012||Virtualization||DCC Ch. 3|
|Virtualization||DCC Ch. 3|
|10-18-2012||Cloud Computing||DCC Ch. 4, 6||Project #3 due
Project #4 (Amazon EC2/S3)
|Cloud Computing||DCC Ch. 4, 6|
|10-25-2012||Understanding the Cost of the Cloud -- Iman Sadooghi|
|Cloud Computing||DCC Ch. 4, 6|
|11-01-2012||Cloud Computing||DCC Ch. 4, 6|
|Cloud Computing||DCC Ch. 4, 6||Project #4 due|
|11-08-2012||Filesystems & Networked file systems||DSPD Ch. 11||Project #5 (Shared FUSE-based filesystem)|
|Parallel Filesystems||External Reading|
|11-15-2012||Distributed Filesystems||External Reading|
|Distributed Filesystems||External Reading|
|11-22-2012||NO CLASS (Thanksgiving Break)|
|Distributed Hash Tables||DCC Ch. 8
|Project #5 due|
|11-29-2012||Exam #2||DCC 3, 4, 6, 8 (DHT), DSPD 7, 8, 11 + External Reading||Exam #2|
We will also use be using the textbook Distributed and Cloud Computing: Clusters, Grids, Clouds, and the Future Internet (DCC) by Kai Hwang, Jack Dongarra & Geoffrey C. Fox (Required). This is the most modern book about distributed systems I have found. Some of the fundemental topics in this book are not covered in enough detail, so for some topics, we will use another textbook, Andrew S. Tanenbaum and Maarten van Steen. “Distributed Systems: Principles and Paradigms” (DSPD), Prentice Hall, 2nd Edition, 2007 (Optional). I encourage you to buy both tetxbooks as they are both excellent, but if you have to choose just one, please buy the first (DCC), and the necesarry optional reading material needed will be provided to the students in class.
Systems Programming (CS351) or Operating Systems (CS450)
There is a course mailing list; you can send mail to the list by sending email to firstname.lastname@example.org. Please see http://datasys.cs.iit.edu/mailman/listinfo/cs495-f12 for more information about the course mailing list.
There will be 6 projects throughout the semester, each worth 10% of the total grade. The projects will be completed in teams of 2 students. The first project will be hands-on, while the others will be primarly programming projects. Some projects will require knowledge of Java, while others will require knowledge of C and/or C++. It is expected that students know the basics of both of these languages.
Projects will be due at 11:59PM on the day of the due date, through BlackBoard. There will be a 15 minute grace period. Any late submissions beyond the grace period will be penalized 10% every day it is late.
There will be 2 exams, one covering the material from the first half of the class, and the second covering the material from the second half. The exams will be individual, but students will be allowed to use their textbooks and any notes they have (on paper). No electronic devices such as phones, eReaders, tables, or laptops will be allowed. Simple calculators can be used.
The exams are scheduled on:
10-09-2012 from 11:25AM - 1:25PM in SB239
11-29-2012 from 11:25AM - 1:25PM in SB239
Please note that they extend for 45 minutes after the usual end of class, but this should not interfere with anyone's other classes due to the lunch period.
There will be no makeup exams.
The following grading scale will be used. The scale will be adjusted downwards based on the overall performance of the entire class. Traditionally, in my classes, the class average score will typically fall in the B-grade range.
A: 90% ~ 100%
B: 80% ~ 89%
C: 70% ~ 79%
D: 60% ~ 69%
E: 0% ~ 59%