Date: Fri, 13 Jul 2001 00:35:04 -0400 From: Bill Moran <wmoran@iowna.com> To: freebsd-stable@freebsd.org Subject: Re: Regarding New FreeBSD BenchMark From Sysadmin Mag (left out a fiew tuning options) Message-ID: <3B4E7A78.5BA85992@iowna.com>
next in thread | raw e-mail | index | archive | help
Some thoughts ... Steve Price wrote: > > On Thu, Jul 12, 2001 at 10:39:15PM -0400, Albert D. Cahalan wrote: > > > > http://www.sysadminmag.com/articles/2001/0108/0108q/0108q_f2.htm This graph shows FreeBSD beating the others in all but 1 category. All three of the top 3 systems begins to fall off at the 128k file boundry, obviously all three systems are tuned/designed to handle large numbers of files that are SMALLER than 128k. There are two questions to raise from this: 1. Why does FreeBSD performance deteriorate FASTER at the 128k mark than windows and Linux? 2. Does this really matter? Are there real world applications where the performance of a server doing nothing but writing large files in mass quantities is the most important thing? If so, then how can FreeBSD's "many large file" performance be improved? > > http://www.sysadminmag.com/articles/2001/0108/0108q/0108q_f3.htm Here, FreeBSD is really only second to Linux overall. considering the fact that the app used to test was probably written specifically for Linux and then ported to FreeBSD, this isn't surprising. Is the source code to this app available? And can obvious coding oopses be located to explain this, or again is there something that can be done to improve high-load performance? And again, how "real world" is this really? How many servers (other than spammers) will be doing nothing but sending the same message to hojillions of people for long periods of time? > > Ah, never mind the untuned Linux box. Check out the untuned > > Windows 2000 box beating the tuned FreeBSD box. I don't see W2k or Linux beating the FreeBSD box except in isolated areas. FreeBSD wins 5 out of 6 scenerios in the Files test and 2 out of 7 in a mail test that is very probably biased toward Linux. In the mail test, FreeBSD seems to pull a respectable second place (overall) and only starts to drop off at extremely high load. > Right and Microsoft has sold the world that their OS is the > best and have plenty of charts to prove it, yet day after > day companies realize that Windows doesn't scale as Microsoft > would have them believe. You can take whatever someone has > the gall to publish as gospel. I'll take what I've seen in > the real world and not some tests by what might be another > biased company looking to make it "gold" with the fad called > Linux. I would lend some credence to these test, since the publishers were nice enough to re-run them when it was pointed out that they were biased. But I still feel the email test is probably biased toward Linux because of the nature of the company that wrote the app used to run the test. As for the evilness of Microsoft or the "fadness" of Linux, I won't comment except to say that both Windows and Linux still work hard to be competetive performance-wise. My personal experience is that nothing out-performs a FreeBSD box, and the tests really do show that. I'd be curious to have some Linux geeks "tune" the Linux install and add that to the comparison sheets. -- It may be that true happiness is nothing more than the ability to *always* know the right thing to say at the right time, whereas true misery is the state of perpetually saying to oneself, "What I *should* have said was..." To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3B4E7A78.5BA85992>