Date: Mon, 26 Feb 2001 17:01:05 -0600 From: Doug Poland <dpoland@polands.org> To: Olivier Cortes <receiver@deep-ocean.net>, questions@freebsd.org Subject: Re: vinum advice sought Message-ID: <20010226170104.A28773@polands.org> In-Reply-To: <20010226234311.A18599@sylgen.deep-ocean.net> References: <20010226161202.A18209@polands.org> <20010226234311.A18599@sylgen.deep-ocean.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, Feb 26, 2001 at 11:43:11PM +0100, Olivier Cortes wrote: > oh oh, 7 SCSI disks ,really ? > > what do you think of : > > * one for /, /tmp, /var > * one for swap and /usr > * 5 in a raid5 array with vinum for /usr/home !! > I was going to put /, /tmp, /var, /usr, and swap on a single 4.3G IDE and all 7 SCSI for a /data mount (/usr/home too, I suppose) Does that make sense? > raid 5 will give good perf. the little partitions are on one disk, they are not too solicited > generally. swap is nearly alone, with /usr. hoping /usr will not be too busy, swap will not > bother you. on /usr/home, you could redirect the squid cache if you are using it. > > if lots of graphics and large files, raid will have at least stripes of 512k. > if remember vinum(8) dealing with 1mb of stripe size if VERY big files. > > could some other people give some other ideas ? > > > olivier > > > PS: > > > my home net example : > -------------------------- > > server: P3 450 / 384Mo / AHA 2940UW / Tekram DC310 > / (128Mo), /tmp (1Go), /usr (1,8Go) and /var (750Mo) on a seagate 4.3 UW (AHA). > /record for cdrecord, and old SCSI2 1go drive on Tekram. > /dev/rsa0 DDS3 on Tekram. > > /dev/vinum/stripe1 : 2 x 30Go IDE stripped with 512k stripe size. > > thanx to the stripping, IDE disks take few CPU time when active :) > > my corporate server: > ----------------------- > > P3 500 / 512Mo / AHA 29160 / 5 x 9Go Seagate LVD > > 1 seagate for /, /tmp, /var /usr > a mirror of 2 stripes of 2 disks for /usr/home, 128Ko stripe size (small office files & big outlook mailboxes...) > > > > On Mon, Feb 26, 2001 at 04:12:03PM -0600, Doug Poland wrote: > > Hello, > > > > I'm building a modest file server for a small, > > non-critical network of M$Win users. Typical > > users will be using M$ Office apps reading and > > writing binary files. Some files will get quite > > large (as in Outlook mail files) and word files > > with lots of graphics. > > > > Unless I'm way off the mark, I'd like to optimize > > for speed so the windows users (PIII 500-900MHz) > > won't be frustrated with delays and be tempted > > to save files on local HD. > > > > This is my config... > > > > Ethernet: 10/100 Intel and 10/100 SMC NICS > > Clients: PIII 500-900MHz (between 5 and 20) > > > > Server: PII 200MHz > > 48MB RAM > > 4.2-RELEASE (soon to be -STABLE) > > Using latest samba > > Host adapter: single AH2940 > > NIC: single 10/100 Intel > > HDs: 3600RPM IBM SCSI-2 (total of 7) > > > > Given this, could someone recommend a vinum > > config that will give the best performance > > possible on this hardware? > > > > I've read and somewhat understand the following > > resources: > > vinum(4) and (8) man pages. > > http://www.vinumvm.org/vinum/Performance-issues.html > > http://www.shub-internet.org/brad/FreeBSD/vinum.html > > http://www.freebsddiary.org/vinum.html > > > > As I understand, there may not be a "best" way > > to configure a vinum system. Since I've never > > done this before, I don't understand the nuances > > between number of spindles, drivers/controller, > > the influences of stripe size, RAID-1 mirroring, > > request size, concurrency, request structure. > > > > I'm not a sysadmin by training so go easy on > > me please :) > > > > -- > > > > Regards, > > Doug > > > > To Unsubscribe: send mail to majordomo@FreeBSD.org > > with "unsubscribe freebsd-questions" in the body of the message > > --- > Olivier Cortes > free software admin -- Regards, Doug 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?20010226170104.A28773>