Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!munnari.OZ.AU!uunet!in1.uu.net!204.71.1.48!newsfeed.internetmci.com!chatta.samart.co.th!news From: Ruediger Koch <rkoch@samart.co.th> Newsgroups: comp.os.linux.networking,comp.unix.bsd.bsdi.misc,comp.unix.bsd.misc,comp.sys.sun.misc Subject: Re: Sparc vs. x86 speed (was Re: Linux vs BSD) Date: Mon, 27 Jan 1997 15:13:19 +0700 Organization: Samart Lines: 69 Message-ID: <32EC639F.3D354CF3@samart.co.th> References: <32DFFEAB.7704@usa.net> <5c155c$p6u@raven.eva.net> <5c19pg$rf6@lynx.dac.neu.edu> <5c58n9$hcb@innocence.interface-business.de> <32E66DE1.7E36AB48@samart.co.th> <5c8b0o$313$1@capsicum.wsrcc.com> <JASON.97Jan23151639@daffodil.cs.odu.edu> NNTP-Posting-Host: ares.samart.co.th Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 3.01Gold (X11; I; Linux 2.0.18 i586) CC: austin@visi.net Xref: euryale.cc.adfa.oz.au comp.os.linux.networking:66213 comp.unix.bsd.bsdi.misc:5738 comp.unix.bsd.misc:2034 comp.sys.sun.misc:28094 Jason C Austin wrote: > > > But what was the end user performance? Did you compare your > mail volume per hour before and after the change? A low load average > usually means you're not getting things to the CPU fast enough to keep > it busy. I've found "PC" type systems tend to have too much CPU for > the rest of the machine, since it looks good on an advertisement. All > you end up with is a hurry up and wait situation which gives you a low > load average. I don't think that our users ceased sending and receiving mail just because we changed our mailserver. We get about 50 new users per week and the mail volume is increasing more than proportional to that :-) > > A load average of 1-2 on a 1 CPU machine is just about where > you want it to be. If it's not, you probably have an I/O bottleneck > of some sort. That's probably one of the greatest nonsense I've ever heard. Reduced load means that I have an I/O problem?! After our change, the response time of the server if you do bash$ telnet yum 25 is a nothing. Before the change it took 1-3 seconds on a 100 MB ethernet! The same is true for POP3. Don't tell me something's wrong with nameserver, the routing or tcp wrapper settings. They're all the same. So much about I/O performance. Doing something like: bash$ ls -l /home > ./homedirs with 6000 userdirs in /home took minutes! Actually, I don't think that this has too much to do with the Sparc hardware, but with it's OS, Solaris. I think I mentioned that we replaced Solaris by Linux on a couple of them, beeing very satisfied with the results (Only one crash till now). I guess, switching to SunOS would have lead to similar results. The Sun hardware is just a little overprized. > > As I said before, CPU is not what makes a fast machine for > most applications. If your application is almost all CPU with low Sounds like a weak CPU makes a good database server?! Always this I/O argument! This depends on your peripherals, of course. You can't compare a Pentium or PPro with 8MB, an ISA 10MBit ethernet card and a slow IDE drive with a $15,000 Sun. That is, however, what is usually done. Compare a $15,000 PC with a $15,000 Sun! > I/O, you probably will be better off with the Pentium machine. If > you're doing database type applications (much more common), the Sun > will leave it in the dust. You need to get the right machine for the > job. Exactly. That why I stopped using Solaris and started to use Linux and FreeBSD on production servers (after I used it for a while as my personal box). We do not have any demanding database apps; for mail, PCs running FreeBSD or Linux are simply better. I don't know if solaris is stable when running Oracle, but I do know that it is not as a mailserver with more than 1000 mails per hour. I've heard of poeple having trouble with 2000 mails per hour on an enterprise 6000. If anyone ever made serious benchmarks and comparisons - they're greatly welcome. R"udiger