Date: Wed, 14 Jan 2004 23:47:16 +1030 From: Phil Kernick <Phil@Kernick.org> To: Ari Suutari <ari@suutari.iki.fi> Cc: Raymond Wiker <Raymond.Wiker@fast.no> Subject: Re: Adaptect raid performance with FreeBSD Message-ID: <4005415C.6090102@Kernick.org> In-Reply-To: <200401141511.30958.ari@suutari.iki.fi> References: <200401141453.50150.ari@suutari.iki.fi> <16389.15817.322098.577889@raw.grenland.fast.no> <200401141511.30958.ari@suutari.iki.fi>
next in thread | previous in thread | raw e-mail | index | archive | help
A note from the dump(8) man page...
Currently, physio(9) slices all requests into chunks of 64 KB. There-
fore, it is impossible to use a larger output block size, so dump will
prevent this from happening.
Same thing likely happens with dd, so bigger than 64k doesn't ever help.
Phil.
Ari Suutari wrote:
> On Wednesday 14 January 2004 15:02, Raymond Wiker wrote:
>
>>Ari Suutari writes:
>> > dd if=/dev/rda1s1a of=/dev/null bs=1m count=100
>> > 100+0 records in
>> > 100+0 records out
>> > 104857600 bytes transferred in 4.193832 secs (25002814 bytes/sec)
>> >
>> > So, I get only about 25MB/s. Shouldn't I be getting something
>> > like 70 MB/s, or even more since there are two disks that
>> > can server read requests ?
>>
>> Have you tried other block sizes? I think you may be able to
>>get better results by going to a lower block size (e.g, 64k instead of
>>1m). Some experimentation will show which block size(s) work best.
>
>
>
> I tried with 32k, 64k, 256k and 512k. Speed is about the same
> with every block size. Block sizes less than 32k seem to give
> even worse performance.
>
> Ari S.
>
--
_-_|\ Phil Kernick E-Mail: Phil@Kernick.org
/ \ ROTFL Enterprises Mobile: 041 61 ROTFL
\_.-*_/
v Humourist, satirist, and probably a few more 'ists to boot!
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4005415C.6090102>