Talk 1: A Data-Oriented Networking Architecture – Scott Shenker Talk 2: RCP: Congestion Control to Minimize Download Time; Nandita Dukkipati

Talk 1:
The current Internet was designed around a host-to-host communication model, but the vast majority of Internet usage today is data access. In this talk, I indulge in the thought experiment of “what if we redesigned the Internet around data access” and present a Data-Oriented Networking Architecture (DONA). DONA includes the use of self-certifying names and a low-level caching/replication infrastructure.

Talk 2:
Rate Control Protocol (RCP) is a new congestion control algorithm designed for fast download times (i.e. aka user response times, or flow-completion times). Whereas other modifications to TCP (e.g. STCP, Fast TCP, XCP) are designed to work for specialized applications that use long-lived flows (scientific applications and supercomputer centers), RCP is designed for the typical flows of typical users in the Internet today. For example, a mid-size flow in the Internet today contains 1000 packets and TCP typically makes them last 10x longer than need-be (XCP is even worse). RCP makes flows finish close to the minimum possible, leading to a perceptible improvement for web users, distributed computing, and distributed file-systems. We believe RCP is the only congestion control algorithm to do this.

The main properties of RCP are:
Typical Internet flows will see 10 times faster download times than TCP and 30 times faster than XCP. Winners are the greater than 90% of sessions that never leave slow-start today. Efficiently uses high bandwidth-delay product networks such as the long haul optical links
Provably stable network independent of link-capacities, round- trip times and number of flows
Flows are easy to police, to ensure they adhere to congestion control (not generally possible with other schemes)
Network operators can give preference (or weighted preference) to some flows/aggregates.

RCP has two components: (1) End-host congestion control layer that sits between IP and TCP/UDP. During introduction, the end-host could adapt by testing for RCP at each end and along the path, falling back to TCP if need-be. (2) Each router maintains a single fair-share rate per link. Each packet carries the rate of the bottleneck link. For each packet, the router compares the two values. If the router’s fair- share rate is smaller, it overwrites the value in the packet. This way, the source learns the fair-share rate of bottleneck link. It is simple, requires a very minor change to switches/routers, requires no per-flow state or per-packet computation.

Speaker Details

Scott Shenker spent his academic youth studying theoretical physics but soon gave up chaos theory for computer science. Continuing to display a remarkably short attention span, his research over the years has wandered from Internet architecture and computer performance modeling to game theory and economics. Unable to hold a steady job, he currently splits his time between the U. C. Berkeley Computer Science Department and the International Computer Science Institute (ICSI).

Nandita Dukkipati received her Bachelor’s degree from Birla Institute of Technology and Science, Pilani, India, and her Master’s degree from the Indian Institute of Science, Bangalore. She is expecting a Ph.D. degree in Electrical Engineering from Stanford University in June 2007. Her research focus is on the design and analysis of various building blocks of network infrastructure, including congestion control, routing, protocol design, router/switch architectures for wired as well as wireless networks. She is particularly interested in building practical networking systems while making use of theoretical tools where applicable.

Date:
Speakers:
Scott Shenker and Nandita Dukkipati
Affiliation:
University of California Berkeley & International Computer Science Institute and Stanford University
    • Portrait of Jeff Running

      Jeff Running