W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > September to December 1995

Re: Keep-Alive Notes

From: Balint Nagy Endre <bne@bne.ind.eunet.hu>
Date: Thu, 12 Oct 1995 18:38:18 +0100 (MET)
To: Dave Kristol <dmk@allegra.att.com>
Cc: http WG <http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com>
Message-Id: <198.bne@bne.ind.eunet.hu>
Dave Kristol writes:
...
> I find it hard to imagine the timeout information to be useful.
Agree again.
I use a 14.4 K dialup line to access the internet. Within hungary a go trough
a 2Mbit radio link, and the international links are 128Kbit lines.
It is easy to saturate any of these links. (e.g. an smtp connection can saturate
my acces line.) I seen one minute RTTs. Measuring the RTT to adjust timeouts
helps nothing, because the load (and RTT) on the lines changes rapidly depending
on the activity of other users.
If proxies are in between, then the load of the proxies adds an other
unpredictable delay. The race condition mentioned by Henrik is unavoidable.

Andrew. (Endre Balint Nagy) <bne@bne.ind.eunet.hu>
Received on Thursday, 12 October 1995 10:51:22 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:31:33 EDT