From: Ruth Ivimey-Cook <Ruth.Ivimey-Cook@ivimey.org>
To: 'Guy' <bugzilla@watkins-home.com>
To: 'Michael Tokarev' <mjt@tls.msk.ru>
To: 'Arshavir Grigorian' <ag@m-cam.com>
Cc: linux-raid@vger.kernel.org
Cc: pgsql-performance@postgresql.org
Subject: Effect of Stripe Size (was Postgres on RAID5)
Date: Tue, 15 Mar 2005 16:17:55 -0000
Message-ID: <20050315161838.B4FC2568C4@svr1.postgresql.org> (raw)
In-Reply-To: <200503142349.j2ENnG510033@www.watkins-home.com>
Folks,
> You said:
> "If your write size is smaller than chunk_size*N (N = number
> of data blocks in a stripe), in order to calculate correct
> parity you have to read data from the remaining drives."
>
> Neil explained it in this message:
> http://marc.theaimsgroup.com/?l=linux-raid&m=108682190730593&w=2
Having read that and the recent posts:
Has anyone done any performance checks on the md code to determine what, if
any, effect the stripe size has on performance? One might suppose the variables
would be stripe size, file size and read vs write. Possibly number of drives in
array, too.
Reason for asking:
When I set up my raid5 array, I chose a stripe of 256K, on the grounds that a
large number of the files on the drive are multi-megabytes (fairly evenly in
the 20MB - 100MB range) and I supposed that a large stripe would speed things
up for those files. Was I right?
Regards,
Ruth
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: pgsql-performance@postgresql.org
Cc: Ruth.Ivimey-Cook@ivimey.org, bugzilla@watkins-home.com, mjt@tls.msk.ru, ag@m-cam.com, linux-raid@vger.kernel.org
Subject: Re: Effect of Stripe Size (was Postgres on RAID5)
In-Reply-To: <20050315161838.B4FC2568C4@svr1.postgresql.org>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox