Internet Congestion Control Research Group (ICCRG)

1
Internet Congestion Control Research Group (ICCRG)

• IETF 69 Chicago July 2007

2
Agenda

• Chairs draft-irtf-iccrg-cc-rfcs (5 minutes)
• Dimitri Papadimitriou ICCRG Open Research
  Problems draft (20 minutes)
• Chairs ICCRG Expert Reviews for the IETF (10
  minutes)
• Murari Sridharan Compound TCP (20 minutes)
• Cesar Marcondes Recent Congestion Control Work
  at UCLA (20 minutes)
• Vijay Subramanian Update on Loss-Tolerant TCP
  Work (20 minutes)
• Dirceu Cavendish Lossless Congestion Control
  (20 minutes)

3
draft-irtf-iccrg-cc-rfcs

• Comments on -01 draft from Sally Floyd, Stephen
  Farrell, Gorry Fairhurst, Lars Eggert, Mark
  Allman
• For -02, in several places we added supporting
  references and/or reworded text
• Added more historic RFCs (970 and 1016)
• Thinking about whether/how to add more on QoS

4
Open Research Issues Document

• Our charter says well document open issues in
  congestion control
• List of issues was gathered at our meeting
  earlier this year and on mailing list
• draft-irtf-iccrg-welzl-congestion-control-open-res
  earch-00
• Dimitri will discuss

5
ICCRG Expert Reviewsfor the IETF

• Basis is ion-tsv-alt-cc
• TCPM, TSVWG, and others will redirect
  experimental congestion control proposals to us
  first
• we analyze under what conditions the proposals
  seem safe to use
• the IETF ultimately decides whether to adopt
  proposals and publish them as RFCs
• several proposals are expected to take this path
  in the coming months toward Experimental RFCs

6
Review Content

• Answer question
• In what environments does this seem safe?
• Make safety statements similar to those in
  QuickStart or HighSpeed RFCs (4782 3649)
• Evaluate each proposal independently
• not a deathmatch competition!
• Some guidelines
• draft-irtf-tmrg-metrics
• draft-ietf-tsvwg-cc-alt

7
Timelines

• Would like to get these done relatively fast
• Need to be responsive to both WGs and authors of
  proposals
• IETF process may take longer
• changes might be iterated back through us
• Workload may increase review times
• Set a goal for 4 months from the time were asked
  to start