home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.sgi.misc
- Path: sparky!uunet!cs.utexas.edu!zaphod.mps.ohio-state.edu!howland.reston.ans.net!spool.mu.edu!sgiblab!sgigate!odin!twilight!zola!anchor!olson
- From: olson@anchor.esd.sgi.com (Dave Olson)
- Subject: Re: Disk striping performance
- Message-ID: <v8cv8ng@zola.esd.sgi.com>
- Sender: news@zola.esd.sgi.com (Net News)
- Organization: Silicon Graphics, Inc. Mountain View, CA
- References: <3980@mutt.mdavcr.mda.ca>
- Date: Sat, 23 Jan 93 19:49:37 GMT
- Lines: 24
-
- In <3980@mutt.mdavcr.mda.ca> young@mdavcr.mda.ca (Shawn Young) writes:
- | I was wondering if anyone had any I/O benchmark results for striped disk sets on
- | SGI boxes, either for just the striped disks, or relative to single disk
- | partitions. I'm looking for something that'll give me an idea of maximum and
- | expected real-world throughput, and a relative measure of increase/decrease in
- | performance with respect to individual disks.
-
- Yes, we have them, but there are way too many variables to give you
- an answer with the info you have provided. The numbers depend a
- great deal on machine type, and raw vs filesystem, controllers used,
- usage patterns, etc. etc.
-
- | Also, I know that the maximum partition size is 8GB, but what happens if I want
- | two partitions: one consisting of 5 1.6GB striped disks, and the other consisting
- | of 4 of the same. Is there any problem with this? What performance penalties, if
- | any, are there?
-
- No problems, no performance penalties per se. You are likely to be
- above the bandwidth of the controllers, so you may get more SCSI
- bus contention, or even system bus on some machines.
- --
- Let no one tell me that silence gives consent, | Dave Olson
- because whoever is silent dissents. | Silicon Graphics, Inc.
- Maria Isabel Barreno | olson@sgi.com
-