Re: [tcpm] TCP tuning

[Joe Touch <touch at ISI.EDU>]

Lars Eggert wrote:
> Hi,
> 
> Jerry Chu has recently started the discussion on whether we need to think about tweaking TCP for the "modern Internet." Just came across another presentation from (AFAICT) another corner of Google that makes similar arguments.
> 
> FYI: http://sites.google.com/a/chromium.org/dev/spdy/An_Argument_For_Changing_TCP_Slow_Start.pdf

I'm curious about the tests described. 200ms seems very high for an RTT.
Even over a cell wireless link (3G), I see latencies that are typically
half that. I wonder what the improvement is for a 50-80ms RTT.

I'm not excited about the prospects of a potential tragedy of the
commons issue (i.e., this works now, but may not work as nicely when
everyone tries this), especially for a mere 50% improvement in latency.

It'd be useful to review the set of proposed mods and claims of impact
for each, though the last time we saw this there seemed to be quite a
few questions about the experiments presented.

Joe

Attachment: signature.asc
Description: OpenPGP digital signature