> If you can shave off 1/3 of the round-trip cost per median object,
> isn't that worth it?
It's more like 15% (1/7) - you burn 2 RTTs in the open and
initial get. It takes another 3-4 'rtts' of transmission. That's
6, or 7 if the smaller initial window is used.
I suppose if the median transfer size was smaller I'd be on stronger ground
but not much. As others have pointed out, server side is what matters.
And both persistent connections and multiple connections defeat this.
How so? multiple I can understand. but persistance has to win surely?
How 'general' are such satelliet connections?
Down here, I suspect over 1/3 of all inbound will be on sat. Is a 1 in 3
chance of doubling the delay/bandwidth product worth the risk?
It might be interesting to review the extent of homebrew assymetrix/sat inbound
deployment. Are we already above critical mass in terms of how many significant
transits outside the US mainstream lie on this kind of path? When I last
looked, there were few online resources which discussed this as deployment
instead of concept, but I think thats less likely to be true now. After all
the tcpsat I-D is really close to the domain of BCP, not just theory isn't it?
And do we now accept proposals "unless they're provably bad"?
(rough consensus and the absence of running, bad code?)
No. Like I said, I've probably over-stated my case. But the aphorism came
to light about 7 iterations of doubling ago, and scale of dispersion of code
and the variance of codebase and platform was smaller. As I believe Mike Lesk
has said a propos UUCP "its easy to push a vacuum aside, moving old code out
is much harder" If a change like doubling initial window was worthwhile,
achieving the rough consensus as a significant number of deployed servers could
prove hard. Especially if the servers are increasingly going to be NT, and
subject to very indirect tuneing methods.
-George
--
George Michaelson | DSTC Pty Ltd
Email: [email protected] | University of Qld 4072
Phone: +61 7 3365 4310 | Australia
Fax: +61 7 3365 4311 | http://www.dstc.edu.au
This archive was generated by hypermail 2b29 : Mon Feb 14 2000 - 16:14:42 EST