From owner-freebsd-hackers Sun Jun 21 16:07:22 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id QAA21397 for freebsd-hackers-outgoing; Sun, 21 Jun 1998 16:07:22 -0700 (PDT) (envelope-from owner-freebsd-hackers@FreeBSD.ORG) Received: from pinky.junction.net (pinky.junction.net [199.166.227.12]) by hub.freebsd.org (8.8.8/8.8.8) with SMTP id QAA21375 for ; Sun, 21 Jun 1998 16:07:13 -0700 (PDT) (envelope-from michael@memra.com) Received: from sidhe.memra.com (sidhe.memra.com [199.166.227.105]) by pinky.junction.net (8.6.12/8.6.12) with ESMTP id QAA00523 for ; Sun, 21 Jun 1998 16:07:01 -0700 Received: from localhost (michael@localhost) by sidhe.memra.com (8.9.0.Beta5/8.9.0.Beta5) with SMTP id QAA29580 for ; Sun, 21 Jun 1998 16:06:50 -0700 (PDT) Date: Sun, 21 Jun 1998 16:06:47 -0700 (PDT) From: Michael Dillon To: freebsd-hackers@FreeBSD.ORG Subject: LSMTP and Unix (fwd) Message-ID: Organization: Memra Communications Inc. MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Why is this benchmark so slow on UNIX filesystems? Will they give out the source to the benchmark so that someone can test FreeBSD's performance and/or figure out if the benchmark is valid? ---------- Forwarded message ---------- Date: Sun, 21 Jun 1998 13:12:38 -0700 From: MJG - inet-access Reply-To: inet-access@mailinglists.org To: "'inet-access@earth.com'" Subject: LSMTP and Unix Since the topic of ListServ and LSMTP came up a bit ago, I thought some of you might find this interesting (please no OS flamewars please!). Matt -----Original Message----- From: Eric Thomas [mailto:eric@LSOFT.COM] Sent: Sunday, June 21, 1998 4:56 AM >Are there any plans to port LSMTP to UNIX? ... The performance >of LSMTP and a UNIX operating system combined would be excellent! LSMTP is being ported to Solaris (SPARC only), AIX and Digital unix. We may add other systems when we are done with these three, but unlike LISTSERV there are serious technical porting issues and we also need to buy a high-end SMP systems with RAID and everything for stress tests. One of our programmers ported it to Linux for fun, but it crashed the thread library in various mystic ways and he gave up for now. Before you volunteer to help, he isn't *allowed* to spend any serious time on this until the other three systems are done :-) Sorry, but market demand dictates priorities. I don't know why people always assume that LSMTP will automatically run faster on unix. Other than pure application code, which is the same on all systems, LSMTP depends heavily on file I/O, network I/O and scheduling. NT has low overhead for all three. In practice the main source of wasted cycles is file I/O. Here are some figures for a LISTSERV benchmark that, while not designed specifically for LSMTP, measures the same kind of I/O that LSMTP does and has been shown to relate to LSMTP performance. Higher figures mean better performance, anything above 50 is very good. There are two different benchmarks, I'll write them down as xx/yy, it is not a ratio but two independent numbers on the same scale, and in practice you want to worry mainly about the second one, so I'll arrange the numbers in that order. First NT (all systems are 4.0 and with the $#%#%# 8.3 DOS compatibility kludge disabled, which is the first thing we do after installing NT on a fresh system): - 166MHz Pentium laptop with toy drive: 34/32 - 300MHz Pentium II with silly 5400rpm IDE drive: 57/54 - 200MHz Pentium Pro with entry-level RAID: 133/121 I have figures for faster CPUs, but they do not improve the results. You can see a difference from a 486 or Pentium to a Pentium II, but that's about as far as the CPU impacts the results. Here are some other systems: - SS20, SunOS 4.1.3, unknown drive: 8/3 - IP22, Irix 5.3, unknown drive: 15/5 - 90MHz Pentium, BSDi 2.0, 5400rpm drive: 7/7 - 90MHz Pentium, Linux 1.2.13, 5400rpm drive: 64/8 - 300MHz UltraSPARC-IIi, Solaris 2.6, 7200rpm drive: 9/9 - RS/6000 (can't figure out the frequency but old), AIX 4.0, undetermined drive: 24/20 - 233MHz Alpha (EV45), Digital unix 3.2, 5400rpm drive: 53/20 - HP 9000/889, HP-UX 10.x, RAID: 75/21 - 400MHz Alpha (EV56), VMS 7.1, RAID, clustered (adds overhead): 44/40 - 533MHz Alpha (EV56), Digital unix 4.0, RAID: 263/108 Obviously the RAID systems have the best numbers, but it is not as clean cut as with NT. Some systems have good numbers even without RAID, some have second rate numbers even with RAID (I have a lot more numbers but I also have work to do :-) ). Note that Digital unix has a revamped file system in 4.0, so you can't compare 4.0 and 3.2 directly (4.0 without RAID would be faster than 53/20). The general idea is that you don't want to use ufs with LSMTP, but even some of the non-ufs systems are slowish. My experience in correlating these numbers with LSMTP performance is that to get good performance without breaking records, you need to score around 30 at least on the first test and ideally on both. All NT systems do that, even the laptop! To break records, you need 100+ ideally, and no less than 50 on the second test. We still manage to break records on VMS, but without boring you with RMS details it is a unique file system with unique issues and in the end you can make it deliver roughly the same performance as a system scoring say 70, plus VMS was engineered for lots of asynchronous I/O and is faster in other areas, which compensates partly for RMS. Even so, the file system is by far the biggest bottleneck on VMS. VMS clearly outperformed NT with the previous generation of processors, now it's about even, and soon NT will take the lead (actually, I think Digital unix will take the lead, but it's too early to be sure, we only have partial lab results). One of the challenges we're facing is how to get the best out of the file system. For Digital unix we can simply tell people to use advfs, which they are probably doing anyway, for systems which only support or include ufs this is going to be a lot more delicate. People always blame us for the shortcomings of the hardware or software they chose without asking our advice ;-), so we know better than to assume it will be easy to explain. I can already hear people bellowing "WHAT??? 'ufs' is the most widely used file system blah blah how can you possibly not magically deliver the best performance on ufs when even NT is able blah blah..." :-( For some reason, people never say "WHAT??? Even NT is faster than ufs? Well, then I need to have a chat with the guy who sold me this box and said ufs was the fastest file system in the world and specifically mentioned being easily 10 times faster than NT!" :-) None of this is going to be an issue if you have a small workload, say 100-200k/day. It is only a problem when you decide to buy a really big box to deliver a whole bunch of mail. This posting will probably have been delivered to 98% of its recipients some 20 sec or so from when I hit "Send," and if it were to take 30 sec instead I am sure you would survive :-) On the other hand, if your large newsletter had to go out in 2h and it took 3h instead, it would be another story. Eric - To unsubscribe from this list, send 'unsubscribe' in the body of an e-mail to 'inet-access-request@mailinglists.org' To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message