Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!news.mel.connect.com.au!munnari.OZ.AU!news.ecn.uoknor.edu!feed1.news.erols.com!cpk-news-hub1.bbnplanet.com!news.bbnplanet.com!su-news-hub1.bbnplanet.com!www.nntp.primenet.com!nntp.primenet.com!news1.best.com!nntp1.best.com!usenet
From: dillon@flea.best.net (Matt Dillon)
Newsgroups: comp.mail.sendmail,comp.mail.smail,comp.unix.bsd.freebsd.misc
Subject: Re: sendmail vs. qmail
Date: 15 Dec 1996 00:27:08 GMT
Organization: BEST Internet Communications, Inc.
Distribution: world
Lines: 39
Message-ID: <58vgks$332@nntp1.best.com>
References: <57tf61$gq7@raven.eva.net> <58h6r1$8ik@ezekiel.eunet.ie> <1996Dec1005.15.53.2968@koobera.math.uic.edu> <SHALUNOV.96Dec13131524@math.wisc.edu>
NNTP-Posting-Host: flea.best.net
Xref: euryale.cc.adfa.oz.au comp.mail.sendmail:35194 comp.mail.smail:2710 comp.unix.bsd.freebsd.misc:32612
:In article <SHALUNOV.96Dec13131524@math.wisc.edu>,
:stanislav shalunov <shalunov@math.wisc.edu> wrote:
:>>>>>> `djb' == D J Bernstein <djb@koobera.math.uic.edu> has written:
:>
:> >> Two extra complete SMTP negotiations and transmissions faster than a
:> >> single RCPT?
:>
:> djb> Profile. Don't speculate. Your experience with serial mailers does
:> djb> not apply to programs handling several jobs at once.
:>
:>I am sorry, but may I humbly ask for confirmation? Let us say that we
:>have, e.g., 5K message (with headers), and, e.g., addresses of normal
:>length, about 30 or 50 bytes.
:>
:>Do you claim that opening 10 connections in parallel and transmitting
:>the same data (except RCPT line which doesn't affect data length much),
:>about 5K, is faster than transmitting the same 5K one time? Do you
:>claim that the worse is connection the more is gain?
If I may interject a point: DNS lookups to failed NS hosts
are notoriously slow, and each attempt DOES generate a number
of packets. But in any medium sized mail system this traffic will
be nothing compared to the nominal mail volume. It's in the
noise, at least on our site. As far as repeated connection attempts go,
non-inclusive of DNS, more recent sendmail releases have both
the MinQueueAge parameter and a host status cache which effectively
removes most of the problem.
As far as openning X connections in parallel goes... there is a huge
problem with doing things that way. You create a potential failure
mode in your mail system that could be hit by anything from a
user sending a multi-recipient mail out to a mail spammer. If you
have any significant mail volume, you just don't do this. Anybody
with significant mail volume knows that the best solution is to have
no failure modes at all (at least none that can bring a machine down),
and barring that to have as few as possible.
-Matt