Date: Mon, 30 Oct 2000 13:09:36 +1030 From: Greg Lehey <grog@lemis.com> To: "Jim C. Nasby" <jim@nasby.net> Cc: freebsd-questions@FreeBSD.ORG Subject: Vinum performance (was: RAID on the cheap) Message-ID: <20001030130936.C37379@wantadilla.lemis.com> In-Reply-To: <39FC34A9.6F37372D@nasby.net>; from jim@nasby.net on Sun, Oct 29, 2000 at 08:31:05AM -0600 References: <200010290700.BAA51057@cs28120-135.houston.rr.com> <39FC34A9.6F37372D@nasby.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sunday, 29 October 2000 at 8:31:05 -0600, Jim C. Nasby wrote: > Zdenko Tomasic wrote: >> >> I would normally prefer scsi setup which would be better, but cost is >> the issue here. What would be recommended as an adequate CPU? PIII-733? > > My brother just asked me the same basic question, so I ran a test: > > tar -cf /dev/null /usr & > > While this was running, I kept an eye on top and couldn't discern any > increase in the cpu useage reported by the vinum process: > > root 18 0.0 0.0 592 8 ?? DLs Wed09PM 0:00.01 vinum: vinum daemon (vinum) The Vinum daemon normally doesn't do anything. It's just there to handle configuration updates and error recovery. > tar was showing ~10% of a CPU. This box is a dual PII-375 with 4 9G > 10kRPM U2W drives in a raid-5 configuration with softupdates enabled. > > Watching the vmstat page of systat, the disks were doing ~100 TPS and > about 600kB/s. This obviously seemed a bit low. tar's not necessarily a good example. Depending on what's on your drives, you may spend most of your time seeking. > Running tar -cf ~/usr.tar /usr shows tar using about 10% of a cpu > (system and interrupt combined are less than 10%, according to top), tps > are down to ~75, but the drives are now doing 1.5-2.5MB/s. Still not > getting the full potential (the drives are seeking their brains out... > /home is a different filesystem that /usr, but they're both raid-5 > volumes on the same set of drives). The speed discrepancy would suggest that you're tarring files of different size. You'd expect this to be slower than to /dev/null. Greg -- When replying to this message, please copy the original recipients. If you don't, I may ignore the reply. For more information, see http://www.lemis.com/questions.html Finger grog@lemis.com for PGP public key See complete headers for address and phone numbers To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20001030130936.C37379>