FW: [Iccrg] ctcp review: big picture issues (3 of 4)

[Murari Sridharan]

Following up on Mark's questions from yesterday's ICCRG meeting.

Mark here is the link I mentioned, the paper is also at the following link

http://www.hep.man.ac.uk/PFLDnet2008/paper/Kun_T%20Final.pdf

Almost all graphs here are due to real congestion (except Fig 7) and since this is improving CTCP behavior under very low buffered links we use buffer sizes as low as 0.6% BDP and losses are truly congestion based as number of flows increase, background traffic is added etc.

If you also look at the SLAC results from real networks those obviously count as well. Also from the wiki if you look at the following presentation there are slides in the end which is trying to look at the stability of the protocol with high level of statistical multiplexing at the bottleneck link. This should give you some idea about worst case behavior of having a very small gamma and also these are results where losses are purely due to congestion. Slides 21/22.

http://research.microsoft.com/users/dthaler/IETF%20-%20Compound%20TCP.pdf

Thanks

-----Original Message-----
From: iccrg-bounces at cs.ucl.ac.uk [mailto:iccrg-bounces at cs.ucl.ac.uk] On Behalf Of Murari Sridharan
Sent: Tuesday, December 04, 2007 3:40 PM
To: mallman at icir.org
Cc: iccrg at cs.ucl.ac.uk
Subject: RE: [Iccrg] ctcp review: big picture issues (3 of 4)

Sorry, that reference is to the ICCRG Wiki

http://www3.tools.ietf.org/group/irtf/trac/wiki/ICCRG_ctcp

The link is gamma parameter tuning.

-----Original Message-----
From: mallman at icir.org [mailto:mallman at icir.org]
Sent: Tuesday, December 04, 2007 3:38 PM
To: Murari Sridharan
Cc: Lachlan Andrew; iccrg at cs.ucl.ac.uk
Subject: Re: [Iccrg] ctcp review: big picture issues (3 of 4)


> If you look at our gamma auto-tuning paper there are a number of such
> experiments.

Can you make that reference a tad more specific?

Thanks,
allman




_______________________________________________
Iccrg mailing list
Iccrg at cs.ucl.ac.uk
http://oakham.cs.ucl.ac.uk/mailman/listinfo/iccrg