Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 20 Oct 2004 11:51:48 +0200
From:      fandino <fandino@ng.fadesa.es>
To:        freebsd-current@freebsd.org
Subject:   Re: FreeBSD 5.3b7and poor ata performance
Message-ID:  <41763534.2060505@ng.fadesa.es>
In-Reply-To: <94275.1098221474@critter.freebsd.dk>
References:  <94275.1098221474@critter.freebsd.dk>

next in thread | previous in thread | raw e-mail | index | archive | help
Hello Poul, nice to know you.

Poul-Henning Kamp wrote:
>>>>Why is a SCSI raid-10 system slower than a plain IDE disk? Something is
>>>>wrong here.
> 
> 
> It has been proven that apples and oranges can indeed be compared,
> (for instance spectroscopically, see ISBN 0-7167-3094-4 page 93),

:-)

> A sequential speed test like the one done here is pointless for
> 99.999% of all disks being used in the world.

I begun this thread because disk performance in two FreeBSD machines was
very low, whilst other operating systems using the same machines were
very superior(x2) to FreeBSD.

I did several tests with bonnie++ and they confirm this low throughput.

A lot of people said "turn off async, softupdates, etc", but nothing
brings FBSD performace to a normal level :-(

It was just a curiosity test sequential reads using raw devices (just to
discard the overhead of filesystem proccesing). It was my down-up test
methodology, if sequential read access with raw disks is bad then sequential
read acces with filesystems can be seriously affected.

> Is your application really writing and reading data only sequentially
> to raw disks ?

not, they read very large files using the filesystem.

> Please do a real and realistic test instead.  One which includes
> seek times, rotational delay and read/write mixes.

I did, please read my posts. If you think I forget something I will be
pleased to do some tests for you.

http://lists.freebsd.org/pipermail/freebsd-current/2004-October/040568.html
http://lists.freebsd.org/pipermail/freebsd-current/2004-October/040740.html

> Or at least realize that the quantity you are measuring is very
> special case that not many people care about in practice and is in
> no way indicative of a general concept like "poor ata performance".

perphars a more appropiate subject will be "poor ata performance with
sequential read/writes" :-?

> There are tools in the ports collection for diskbenchmarking, please
> use them rather than come up with some half-assed home-brew stuff.

yes, I run dd, bonnie++, bonnie and reports are very clear, there is
a problem with performance and the ata system (I think).

Regards.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?41763534.2060505>