Re: [PATCHES RFC 2/2] Optional hash table improvements
Date: Wed, 2 May 2007 20:49:28 +0200
On 5/2/07, Daniel Stenberg <daniel_at_haxx.se> wrote:
> >> I mean, if someone goes mad and tries 100,000 connections they probably
> >> will see that upping this size even more will be very beneficial to them...
> > Actually, we were really planning doing that in a couple of months... and
> > our madness is limited only by some more urgent features/issues and absence
> > of appropriate HW.
> What's the largest number of connections you've "done" so far?
15K against local lighttpd with an ancient home Pentium-4 3.0 GHz and
1GB of a slow
unchecked memory with a fine tuning of some kernel params as here:
The main issues to reach 50-100 K may be a strong CPU consumption
by CreateConnection (still to investigate - could be a memory allocation
issue), enough memory - 4-8 GB of a fast checked memory on a good bus
with a good dual CPU server, like Supermicro or Tyan AMD Opteron.
A server side able to run 50-100 K connections may be lighttpd (epoll mode
with fd-max increased) with another piece of a good HW, where a single CPU
server with 2GB memory might be enough.
-- Sincerely, Robert Iakobashvili, coroberti %x40 gmail %x2e com ........................................................... http://curl-loader.sourceforge.net A web testing and traffic generation tool.Received on 2007-05-02