Home

Program

Venue

Sponsors

The symposium consists of three parts:

• Architecture Parade
  We have asked distinguished researchers to present and critically review past design choices they were involved with.
  These keynote speeches will start Monday afternoon (March 16) and extend until Wednesday (March 18). The kenote speeches will always be followed by a moderated discussion (reconceptualization) with the audience.
• Research projects
  We have invited individual researchers to contribute their analysis and views in short presentations that will be grouped and placed between the "Arch Parade" interventions ("back to the future" slots).
• Small discussion groups
  Participants will be actively supported to gather in smaller groups and discuss important architecture topics of their choice. (Monday to Thursday). Collected insights will be documented and discussed as much as possible during the event, especially on Thursday. The outcome will be documented to the whole networking community.

Scientific Aim

From a scientific point of view, the Internet is often seen as an extremely positive development: Among other benefits, it changed the way academia works, creating a global and easily accessible platform for scientific collaboration, and created a wonderful showcase for networking technology.

At the same time, the Internet has dominated the computer science part of networking over the last decade, to the extent that little attention has been paid to alternative approaches. Today we still lack a comprehensive theory of networking; progress in this area has been rather slow in the past decade. Meanwhile, the Internet has grown out of understanding and control. This has led to numerous (network architecture) proposals with each new PhD generation, many of which re-introduce previously discarded approaches.

Recently there are signs of change: we are seeing a renaissance of major networking research initiatives that recognize the need for an overhaul of the Internet all the way to its foundation. In this context, the NetArch 2009 conference aims to gather and record the field's understanding of networking theory, and to create a platform for discussing directions we should take to improve our understanding of computer networking.

The "why" of Networking

In networking textbooks, we find a lot about "what" a (Internet) network is and "how" it works, but very little about the "why". This tends to shroud the inventions in mystery and mask the pragmatic decisions that had to be taken in order to get things working. The NetArch 2009 event shall therefore concentrate on unravelling and documenting the rationale, by inviting the first generation of Internet researchers and engineers to reconstruct (and perhaps deconstruct) their mindset of 40 years ago, as well as today. This shall also include researchers whose proposals at that time were not successfully adopted—for example SNA, DecNet, OSI or Xerox. To this end, we will run a 3-day "Architecture Parade" for which key networking persons will document their (retrospective) insights into networking principles.

Key questions of interest are for example:

• How much of today's architecture is an artifact of 1970's technology? (Example: header size and MTU have stayed constant while bandwidths have increased by many orders of magnitude.)
• What should be "named" (or "identified") in a network system? To what entities or abstractions should we assign identifiers? Do we even need identifiers, or are attributes sufficient? At what level in the "stack"?
• Is hop-by-hop routing/forwarding the right approach? Have we adequately explored the space of alternatives? Is DTN a natural extension of store-and-forward?
• What is the right "paradigm" to support mobility?
• Which abandoned technology choice should be reconsidered and re-researched? (PIP, Delta-T, Nimrod etc)
• What aspects of the TCP/IP architecture are "right", and which ones would be done differently and (most importantly) why?

The NetArch 2009 symposium concentrates on the architecture, rather than specific networking problems like routing or congestion control, for three reasons: First, networking architecture is about setting strong boundaries on what techniques can and must be put into operation—and in particular, must be built into the infrastructure—for a network to function effectively. Second, architectural contributions have not had a prominent place in the existing publication channels. And third, we are witnessing an upsurge of research in network architecture, which can only benefit from mutual exposure, direct comparison, and interchange of ideas.

The split horizons meet

Thanks to the success of the Internet, the networking community has become large and quite diverse. A side effect of this success is that it becomes increasingly difficult to track all relevant developments and projects—especially for such a broad topic as network architecture design. Yet it is crucial that any attempt at a new Internet incorporate all the latest applicable research to the greatest possible extent, lest the result be multiple, incompatible network layer designs, ultimately requiring "down-conversion" to the least common denominator in order to achieve full connectivity.

To encourage cross-fertilization and confrontation of ideas, the NetArch 2009 symposium will include workshops and presentations focusing on all major architecture projects and initiatives going on worldwide ("back to the future" slots). This is expected to promote synergies and collaborations, and could open up new exciting research directions where, for example, new inter-networking technologies could be designed in parallel to the new network architectures currently being developed.

• Overall
• Scientific aim
• The "why" of networking
• Split horizons meet