CSE 570. Reinventing the Internet

Course Description

The Internet is one of the great technology success stories of the last twenty years. As it has grown, it has been used in ways that the original designers never envisioned and for which the core protocols were never designed. New needs have been met through a variety of ad hoc mechanisms, which have never been integrated into the network architecture, and which, in many cases, violate some of the Internet's central architectural principles. This course will study how the Internet has evolved to meet changing needs and will explore the technical issues that have arisen, as a result of this evolution. It will also explore how the Internet might be restructured to address these issues. The course will be based primarily on published research papers, with reference to standard texts and RFCs for background material.

Administrivia (Fall 2006)

The class will be conducted using a combination of conventional lectures and student-led presentations of selected papers from the research literature. In particular, each Monday class session will involve student presentations of two different papers, and each Wednesday session will involve a lecture from the instructor. The student presentations will be limited to 25 minutes, to allow time for discussion. This means you will need to focus on presenting the most important elements of the paper, rather than presenting every last detail. Student presentations must be mailed to the instructor by 6:00 pm the evening before the presentation. They will be made available on the web site, so that other students can review them before class.

For each of the student-led presentations, several other students will be designated to write critical reviews of the papers. These will be due on the Friday before the paper is to be presented at 6:00 pm, and must be posted as articles to a blog that has been setup for the class. Reviews should be one to two pages long. I suggest you look at the public reviews from recent HotNets conferences and this year's SIGCOMM conference for examples of the kind of reviews I am expecting. However, that I advise you not to read online reviews of any paper that you will be reviewing yourself.

The papers that will be used for the student presentations are listed at the bottom of this page. To sign up to present or review a particular paper, please use this form (there is also a link to the form in the left margin). Each student is expected to give two presentations over the course of the semester (at least one must be before the middle of October) and review six papers (say two per month).

Each week, several students will also prepare short essays on one of several selected topics. Essays should be 3-4 pages long. These are also to be posted to the blog and will be due on Sundays at 6:00 pm. Each student will do three essays over the course of the semester (say one per month). Signups for essays will also be done using the scheduling form.

There will be a final paper due at the end of the semester. This should take the form of a proposal for a novel network architecture or architectural feature. The Hotnets conference is a good source of examples of the kind of work I am looking for here. Each student will also give a short presentation of their final paper. These will be done during finals week.

Students are also expected to post comments to the blog, addressing issues raised in the papers read, the presentations, the reviews, the essays or the comments from other students. Comments to the blog will be evaluated as part of the course grade.

The blog is called Reinventing the Internet. To register on the blog, click on the "Login" link, in the Meta section at the right and then select the "Register" link in the dialog box. Please use your full name as your user name. Once you have registered, you will be able to write articles that will be saved in the blog database and edit them later. You won't be able to post them to actually appear on the blog, but I will do that whenever the deadline for the given review or essay has passed. Your may also post comments to any of the articles on the blog.

In addition to the blog, there is a usenet newsgroup (wu.cse.class.570) that you can access through newsreader.wustl.edu. Please use this for course-specific questions, discussion that is not directly related to the papers, reviews or essays. I will also use it for announcements, so please check it regularly.

Academic integrity. In a course like this, there may be a temptation to "borrow" from existing online sources. Resist such temptations. Instances of plagiarism will be taken very seriously (a failing grade in the course and referral to the academic integrity committee). You may quote others' work if you cite it appropriately, and you may incorporate selected figures from other sources, but only if you cite them.

Lecture Schedule

  1. Network Architecture (2 lectures).
    Read K&R (3rd edition) pages 430-476, 314-383, 196-200, 228-253, 146-163, 123-135 and NewArch Final Report and The material from K&R should be review, but I do urge you to read either the pages indicated in K&R or comparable material from other textbooks (Stallings or Peterson/Davie are both fine for this purpose). The K&R page numbers above are from the third edition (the corresponding 1st edition page numbers are: 415-441, 300-341, 207-231, 124-153, 177-182). If you are using another book, the subject material to review is: Ethernet (including hubs, bridges and switches), IP protocols (including IP, ICMP, RIP, OSPF, BGP, TCP, UDP, DNS), operation of IP routers (including lookups and queueing), basic socket programming.
  2. Overlays and Network Virtualization (4 lectures).
    Read Globally Distributed Content Delivery, Democratizing content publication with Coral, PlanetLab Architecture, GENI: Design Principles and GENI Conceptual Design.
  3. Diversifying the Internet (2 lectures).
    Diversifying the Internet, How to Lease the Internet in Your Spare Time, A Proposed Architecture for the GENI Backbone Platform.
  4. Building Metanetworks (2 lectures).
    SANDS: Specialized Active Networking for Distributed Simulation, A DoS limiting Network Architecture.
  5. Diversifying the Access (2 lectures).
  6. Substrate Control Metanet (1 lecture).
  7. Substrate Routers and Metarouters (2 lectures).
  8. Metanet Backbone Configuration (1 lecture).

Student Presentation Schedule

Date Presenter Paper Reviewers
9-11-2006 Charlie Wiseman Addressing Reality: An Architectural Response to Real-World Demands on the Evolving Internet. Andrew Wan, Amy Freestone
9-11-2006 Brandon Heller Towards an Evolvable Internet Architecture. Travis Keshav, Paul Moceri
9-18-2006 Amy Freestone Invariants A New Design Methodology for Network Architectures. Andrew Wan, Michela Becchi, Michael Roche
9-18-2006 Michael Roche A Clean Slate 4D Approach to Network Control and Management. Harri Thorvaldsson, Charlie Wiseman
9-25-2006 Mike Wilson Overlay-Friendly Native Network: A Contradiction in Terms?. Paul Moceri, Niarcas Jeffrey, Nuzhet Atay
9-25-2006 Harri Thorvaldsson ACMS: The Akamai Configuration Management System. Travis Keshav, Brandon Heller
10-2-2006 Andrew Wan OpenDHT: A Public DHT Service and Its Uses. Charlie Wiseman, Niarcas Jeffrey, Michela Becchi, Nuzhet Atay
10-2-2006 Travis Keshav Colyseus: A Distributed Architecture for Online Multiplayer Games. Harri Thorvaldsson, Brandon Heller, Sailesh Kumar, Amy Freestone
10-9-2006 Paul Moceri Network Capabilities: The Good, the Bad and the Ugly. Charlie Wiseman, Mike Wilson, Sailesh Kumar
10-9-2006 Sailesh Kumar SybilGuard: Defending against Sybil Attacks via Social Networks. Harri Thorvaldsson, Travis Keshav, Michela Becchi
10-16-2006 Nuzhet Atay Understanding the Network-level Behavior of Spammers. Mike Wilson, Niarcas Jeffrey, Paul Moceri
10-16-2006 Michela Becchi Off by Default!. Harri Thorvaldsson, Brandon Heller, Charlie Wiseman, Sailesh Kumar
10-23-2006 Amy Freestone An end-to-end approach to host mobility. Travis Keshav, Niarcas Jeffrey, Michael Roche
10-23-2006 Michael Roche A Network Architecture for Heterogeneous Mobile Computing. Andrew Wan
10-30-2006 Charlie Wiseman Geographic Routing Made Practical. Andrew Wan, Mike Wilson, Nuzhet Atay, Brandon Heller
10-30-2006 Niarcas Jeffrey Architecture and Algorithms for Scalable Mobile QoS. Amy Freestone, Michela Becchi
11-6-2006 Paul Moceri Internet Indirection Infrastructure. Brandon Heller, Niarcas Jeffrey, Nuzhet Atay, Michela Becchi
11-6-2006 Harri Thorvaldsson The Design and Implementation of a Next Generation Name Service for the Internet. Andrew Wan, Charlie Wiseman, Mike Wilson, Michael Roche
11-13-2006 Travis Keshav Cashmere: Resilient Anonymous Routing. Harri Thorvaldsson, Mike Wilson, Amy Freestone, Sailesh Kumar
11-13-2006 Andrew Wan ROFL: Routing on Flat Labels. Travis Keshav, Charlie Wiseman, Nuzhet Atay, Sailesh Kumar
11-20-2006 Brandon Heller Enabling Contribution Awareness in an Overlay Broadcasting System. Amy Freestone, Nuzhet Atay, Paul Moceri
11-20-2006 Michela Becchi OverQoS: An Overlay Based Architecture for Enhancing Internet QoS. Harri Thorvaldsson, Sailesh Kumar, Mike Wilson, Michael Roche
11-27-2006 Nuzhet Atay Revisiting IP Multicast. Michela Becchi, Michael Roche
11-27-2006 Sailesh Kumar OASIS: Anycast for Any Service. Travis Keshav, Paul Moceri
12-4-2006 Mike Wilson Quorum: Flexible Quality of Service for Internet Services. Andrew Wan, Niarcas Jeffrey, Amy Freestone
12-4-2006 Niarcas Jeffrey Designing DCCP: Congestion Control Without Reliability. Brandon Heller, Paul Moceri, Michael Roche

Essay Schedule

Date Authors Topic
9-10-2006 Mike Wilson Ethernet - a brilliant example of network architecture and evolution
9-10-2006 Andrew Wan Ethernet - a modest technology gone bad
9-17-2006 Niarcas Jeffrey The Internet protocol suite - a triumph of thoughtful design and engineering
9-17-2006 Travis Keshav How the Internet succeeded in spite of IP
9-24-2006 Harri Thorvaldsson Cross-layer transparency - the key to matching network performance to application needs
9-24-2006 Charlie Wiseman Why well-defined interfaces are essential for enabling networks to evolve and survive
10-1-2006 Paul Moceri Telepresence in real and virtual worlds - the next big thing in networked applications
10-1-2006 Andrew Wan Why network (in)security will remain the Achilles heel of peer-to-peer and ad hoc nets
10-8-2006 Nuzhet Atay, Amy Freestone How overlay networks will make IP irrelevant without actually killing it
10-8-2006 Brandon Heller The absurdity of expecting unstructured overlays to improve application performance
10-15-2006 Sailesh Kumar The essential role of networks in denial of service attack defense
10-15-2006 Travis Keshav The futility of in-network DDoS defenses - it's the computer systems, stupid
10-22-2006 Niarcas Jeffrey, Brandon Heller Re-thinking networks for mobility - static endpoints are just a (rare) special case
10-22-2006 Michela Becchi The role of the wired infrastructure in wireless nets for enabling services on power-poor devices
10-29-2006 Paul Moceri, Mike Wilson The case for strong authentication of network traffic
10-29-2006 Charlie Wiseman, Harri Thorvaldsson You'll have to pry address spoofing from my cold, dead, spam-stained hands
11-5-2006 Amy Freestone, Nuzhet Atay Making networks smarter - the key to reducing administrative and maintenance costs
11-5-2006 Mike Wilson Lord save me from so-called smart networks - building a house of cards on a foundation of sand
11-12-2006 Niarcas Jeffrey, Michael Roche Why multi-provider networks with no central control are bad for users
11-12-2006 Andrew Wan, Nuzhet Atay How competition among network providers can produce better service for users
11-19-2006 Travis Keshav, Harri Thorvaldsson The trouble with Internet naming and the growing abuse of DNS
11-19-2006 Charlie Wiseman, Michael Roche The case for a consistent global naming system for all networked objects - real and virtual
11-26-2006 Paul Moceri, Michela Becchi, Sailesh Kumar Geographic addressing in WANs to simplify routing and enable new services
11-26-2006 Brandon Heller Why geographic addressing has no place in WANs (and maybe not LANs)
12-3-2006 Michela Becchi, Michael Roche The case for making multicast a first-class service in the Internet
12-3-2006 Sailesh Kumar Why multicast is irrelevant to the Internet

Final Project Schedule

Date Speaker
12-6-2006 Charlie Wiseman
12-6-2006 Harri Thorvaldsson
12-11-2006 Mike Wilson
12-11-2006 Sailesh Kumar
12-11-2006 Paul Moceri
12-18-2006 Niarcas Jeffrey
12-18-2006 Michael Roche
12-18-2006 Brandon Heller
12-18-2006 Michela Becchi
12-18-2006 Nuzhet Atay
12-18-2006 Travis Keshav