Any clues to source of this delay?

Markus Stenberg mstenber at cc.Helsinki.FI
Thu Aug 5 06:12:57 CEST 1999


"Robin Boerdijk" <robin.boerdijk at nl.origin-it.com> writes:
> > Oh, sorry, wasn't being explicit; it isn't slow start, as the ceiling is
> on
> > _single_ connection, however long it stays up. By opening multiple
> > connections, the server's total handling capacity goes up but single
> > connections' speed stays constant.
> At the risk of being completely wrong: your problem might be caused by the
> fact that TCP has a sliding window of 64Kb. This means that there can never
> be more than 64Kb of outstanding, unacknowledged data on a single TCP
> connection. When TCP reaches that limit, it will stop sending data until an
> acknowledgement from the receiver makes the window 'slide' to the right. The
> larger the size of the packets you sent, the sooner the windows fills up and
> the less rountrips per second you will get.

Data transferred was minimal.. _apparently_ it was mainly caused by delayed
ack, although, I doubt I'll ever know for sure (too lazy to reread RFC's;
delayed ack would fit the picture though, or possibly slow start without
window size growing).

> Hoping this guess is not totally off,
> 
> Robin.

-Markus

-- 

	Only in the silence voice, Only in the darkness light, 
	Only in dying life; The hawk's bright flight on the empty sky
		-- Ursula Le Guin / (the) Earthsea Quartet




More information about the Python-list mailing list