IRTF Update-November 2010

By: Aaron Falk

Date: October 1, 2010

line break image

This is a status report on the Internet Research Task Force (IRTF) and an introduction to the Transport Modeling Research Group based on the report presented at IETF 78 in Maastricht, Netherlands.

IRTF Status

Eight of the 13 IRTF research groups (RGs) met at IETF 78. Eleven research groups are active at this time, either because they are holding meetings or because they have an active mailing list. The Internet Architecture Board met with the chairs of the recently formed Virtual Networks RG to discuss the group’s charter, activities, and plans.

No new IRTF-stream RFCs have been published since IETF 77. The Internet Advisory Oversight Council is sponsoring an effort to modify the IETF Draft Tracker in order to show progress of non-IETF-stream Internet–Drafts, including IRTF-stream documents. ID draft-hoffman-alt-streams-tracker collects the review and approval states of IRTF documents and other tracker modifications that are required.

A new mailing list has been created for public discussion of IRTF-related topics, such as new RG proposals. The list address is [email protected]. (This is a corrected address from one given in a prior status report.)

Several topics are under discussion to become candidates for new IRTF RGs, including cloud computing, economics and policy, Internet of things, machine learning and communication systems, and social networks.

In spring 2011, I will be stepping down from the position of IRTF chair, which I have held for six years. An effort is under way to identify candidates for the position. If you would like to be considered for the position, please contact Internet Architecture Board (IAB) chair Olaf Kolkman ([email protected]) or me[email protected]). The IRTF chair is approved by the IAB.

Introduction to the Transport Modelling RG

The goal of the Transport Modeling RG is to improve methodologies for evaluating transport protocols. The charter sets out to produce a variety of deliverables, including (1) a survey of models being used in simulations, analysis, and experiments for the evaluation of transport protocols; (2) a broad set of simulation test suites; and (3) a slate of recommendations for test suites that can be used as experiments in test beds.

A transport model is a set of assumptions about network and traffic conditions implicit in protocol evaluation. It addresses the topology, round-trip times, flow arrivals and durations, and flow greediness. Note that it need not be a mathematical model.

A recent discussion in the RG covered evaluation of the effect of faster flow start-up. There has been a proposal in the IETF to increase the TCP initial window, and some of the measurements taken at data centres show improvement in flow completion times. The RG is considering how one would assess degradation to other users from this more aggressive approach, with special consideration of users with slow connections.

Other IETF and IRTF efforts involve transport protocols and congestion control. Therefore, it is useful to be clear about what the Transport Modeling RG is not chartered to do. Specifically, the group does not discuss either the design of new congestion control mechanisms or modifications to existing congestion control mechanisms (except in terms of models needed for the evaluation of such mechanisms). Neither does the group produce evaluations of specific congestion control mechanisms.

The RG has produced a variety of outputs, including:

  • Metrics for the Evaluation of Congestion Control Mechanisms, RFC 5166
  • Tools for the Evaluation of Simulation and Testbed Scenarios, draft-irtf-tmrg-tools
  • Common TCP Evaluation Suite, draft-irtf-tmrg-tests
  • An NS2 TCP Evaluation Tool Suite, draft-irtf-tmrg-ns2-tcp-tool

Future work will include consideration of metrics for evaluating admission control systems and documenting known so-called corner cases in transport modeling.

This article was posted on 31 January 2011