Date: Wed, 6 Sep 2000 12:04:57 +0200 From: Brad Knowles <blk@skynet.be> To: Mike Tancsa <mike@sentex.net>, Greg Lehey <grog@lemis.com> Cc: stable@FreeBSD.ORG Subject: Re: Vinum (was: RAID) Message-ID: <v04220802b5dbc50700be@[195.238.1.121]> In-Reply-To: <4.2.2.20000906000615.03226880@mail.sentex.net> References: <4.3.2.7.0.20000905165313.037652c0@marble.sentex.ca> <4.2.2.20000903213318.074039f0@mail.sentex.net> <4.2.2.20000903213318.074039f0@mail.sentex.net> <00090310375700.05988@www.runapplications.com> <4.2.2.20000904120154.07455bd0@mail.sentex.net> <20000905101653.A49732@wantadilla.lemis.com> <4.2.2.20000904204407.033e2920@mail.sentex.net> <20000905115126.A14470@myhakas.matti.ee> <4.3.2.7.0.20000905090105.053a13f0@marble.sentex.ca> <20000905155032.A27690@futuresouth.com> <4.3.2.7.0.20000905165313.037652c0@marble.sentex.ca> <4.2.2.20000906000615.03226880@mail.sentex.net>
next in thread | previous in thread | raw e-mail | index | archive | help
At 12:29 AM -0400 2000/9/6, Mike Tancsa wrote: > I will certainly take a look at it. However, for now, I have been > using a number of bonnie processes simultaneously. The box I am > building is going to be just a pop3 server. So at any given time, a > number of popper processes running scanning through a users mailbox > to see how many messages they have seems to be well enough > approximated by a number of disk intensive processes like bonnie at once. For simulating the performance of mail-related disk subsystems, so far I have been pretty happy with what I've been doing with postmark. See my postmark benchmarking page at <http://www.shub-internet.org/brad/FreeBSD/postmark.html>. And now that the SmartRAID V/VI drivers are available for FreeBSD 4.2, I will be repeating my previous vinum/SmartRAID/rawio benchmarks (see <http://www.shub-internet.org/brad/FreeBSD/vinum.html>) to see how things have changed. On our POP3 mail systems, we're seeing average message sizes around 70KB, so I will be doing further bonnie++ 1.00b (see <http://www.coker.com.au/bonnie++/>) tests averaging about that, of course using multiple processes in parallel, and making sure that I move at least multiple GB of data per test to try and make sure that we eliminate as much buffer cache effect as we can. I'm also going to be testing Sun Solaris UFS against UFS Logging (both DiskSuite trans meta devices as well as using the "logging" keyword in the mount options) against Veritas VxVM and VxFS, and in single disk, mirrored pairs, striped pairs, mirrored+striped pairs (four disks), and four pure striped disks. > The only problem I have run into so far is creating a 3 disk RAID5 > array. Is this not possible ? While possible (three disks is the minimum number required to make a RAID-5 volume), it wastes 1/3 your disk space, it is very costly in terms of I/O, and generally is not that great of an idea. > Am I better off in the end going with 4 drives in a RAID 10 config instead ? If that is possible, in my experience you will find that it will perform much, much better, and it should be more reliable. Myself, I'd be inclined to throw in a fifth disk as a hot spare, and run the remaining four in RAID 0+1 mode (of course, if at all possible you want to stripe the mirrors and not mirror the stripes, since if you do the former and you lose a disk, then only that submirror is degraded and the stripe is fine, while if you do the latter, then the entire stripe is degraded which then causes the mirror to be degraded too). -- These are my opinions -- not to be taken as official Skynet policy ====================================================================== Brad Knowles, <blk@skynet.be> || Belgacom Skynet SA/NV Systems Architect, Mail/News/FTP/Proxy Admin || Rue Colonel Bourg, 124 Phone/Fax: +32-2-706.13.11/12.49 || B-1140 Brussels http://www.skynet.be || Belgium "They that can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety." -Benjamin Franklin, Historical Review of Pennsylvania. 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?v04220802b5dbc50700be>