CSE 561: Computer Networks

CSE Home

About Us

Search

Contact Info

Networked systems are appearing at all granularities, from planetary scale web services such as Google and Akamai, to distributed databases for managing large corporations, to massively parallel multiplayer games, to large scale sensor networks. In each case, there is a need for a deep understanding of fundamental principles of networked systems in order to achieve optimal and secure execution. It is within this context, we will explore the following types of questions in this graduate level course on networked systems.

• How do we design and build networked systems when there are only a handful of "computers" in the world (namely, Google, Amazon, and Microsoft)? What are the implications of having most of the Internet traffic being sent and received from these data centers?
• Almost every user device on the other side of these communications is a mobile device. Is the Internet designed well for this usage pattern? If we were to design from scratch, would we come up with a different design for the basic communication protocols (e.g., TCP), so as to better suit the mobile setting? Do we need more advanced statistical techniques (e.g., from machine learning) to deal with the vagaries of these networks?
• Networked computing is increasingly integral to all aspects of society. Countering this trend, networked systems face challenges at every level of the protocol stack, e.g., the frequent misconfigurations and ``brownouts'' observed in the Internet, the disruption of service by both overt state actors (e.g., censorship) and non-state actors (e.g., DDoS attacks by botnets and traffic filtering by ISPs), etc. How do we build a more resilient Internet given these challenges?
• The distinctions between hardware and software, network routers and end-host computers, system code and application-level programs are fast disappearing. In this context, shouldn't we be developing programming language solutions for both communicating and managing networks? In other words, what are the limits of software-defined networking?
• Cloud-based services can improve reliability and support ubiquitous access. Yet, they face considerable challenges in the form of economic, performance, and security issues. Interestingly, peer-to-peer systems have a complementary set of strengths and weaknesses suggesting the applicability of a hybrid approach. What are the strengths and limits of these two contrasting forms of computing? Can we build a peer-to-peer version of Facebook or Twitter?

• You will read one or two papers per lecture. The reading list is a mixture of classical research material and more recent papers that focus on emerging trends. For each paper, you will craft a blog entry. This will constitute about 15% of the total grade for the course.

• There will be two quizzes covering material from the papers. This will constitute about 30% of the total grade.

• The best way to learn how to build networked systems is by practice, and so the core of the course is a substantial distributed system design and implementation project. You will form teams of two or three, and each team will pick a research project to do for the quarter. There will be intermediate milestones associated with the project, as well as a final report and presentation at the end of the quarter. This will constitute about 55% of the grade for the course.

Administrivia

Lectures: Mondays and Wednesdays, Noon-1:20pm.

Mailing list: When you register for the course, you'll automatically be added to the class mailing list (cse561a_wi12@uw.edu). To manage your subscription, visit the mailing list web page. You've been subscribed using your u.washington.edu email address. But, you can modify your subscription to use an email address of your choice. Note that you can only post to the mailing list from your subscribed email address.

Announcements:

• [Jan 5] The mailing list is activated.
• [Dec 26] The first lecture for the course will be on Wednesday Jan 4th, as scheduled.

Discussion Board

Here's the link to the class discussion board:

https://catalyst.uw.edu/gopost/board/arvindk/25378/

The discussion board will be used to post announcements, questions, and other administrivia. For each lecture other than the first, we will create a discussion thread for the papers that we will discuss in the class. You are required to add a unique comment to the discussion for each of the threads by 9am on the day of the class. Please keep the entries short: they can be anything that provides insight into the paper, e.g., a summary, the broader context for the work, a question about some aspect of the paper, an answer to someone else's question, a methodological flaw, or a pointer to related work not described by the paper.

	Computer Science & Engineering Box 352350, University of Washington Seattle, WA  98195-2350 (206) 543-1695 [comments to arvind] Privacy policy and terms of use