*BSD News Article 74049


Return to BSD News archive

Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!munnari.OZ.AU!news.ecn.uoknor.edu!news.ysu.edu!usenet.ins.cwru.edu!agate!howland.reston.ans.net!swrinde!news-res.gsl.net!news.gsl.net!news.mathworks.com!newsfeed.internetmci.com!pier2.bayarea.net!baygate.bayarea.net!thorpej
From: thorpej@baygate.bayarea.net (Jason R. Thorpe)
Newsgroups: comp.os.linux.networking,comp.unix.bsd.netbsd.misc,comp.unix.bsd.freebsd.misc
Subject: Re: TCP latency
Date: 18 Jul 1996 00:40:05 GMT
Organization: George's NetBSD answer man
Lines: 15
Message-ID: <4sk155$721@pier2.bayarea.net>
References: <4paedl$4bm@engnews2.eng.sun.com> <x74tnfn35s.fsf@oberon.di.fc.ul.pt> <4s33mj$fv2@innocence.interface-business.de> <4sbrcr$rqd@enomem.lkg.dec.com>
NNTP-Posting-Host: baygate.bayarea.net
Xref: euryale.cc.adfa.oz.au comp.os.linux.networking:45590 comp.unix.bsd.netbsd.misc:4100 comp.unix.bsd.freebsd.misc:23862

In article <4sbrcr$rqd@enomem.lkg.dec.com>,
Matt Thomas <matt@3am-software.com> wrote:

>The second problem is that TCP traverses all of its PCBs looking for
>timeouts.  However, only a small number of PCBs typically have timeouts
>pending.  When supporting large numbers of connections (>20,000),
>the TCP timeout routines can consume a great deal of CPU time when
>they are actually doing very little.

NetBSD has partially addressed this.  The PCBs are now hashed.  This
was done to address some problems on the NetBSD.ORG mail server.
It helped tremendously.
-- 

	Jason R. Thorpe <thorpej@bayarea.net>