Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 19 May 2009 07:06:30 +0200
From:      Erik Trulsson <ertr1013@student.uu.se>
To:        Yuri <yuri@rawbw.com>
Cc:        Glen Barber <glen.j.barber@gmail.com>, freebsd-questions@freebsd.org
Subject:   Re: What is the highest hard drive read/write speed you were able to  achieve by entire disk mirroring or striping?
Message-ID:  <20090519050630.GA16959@owl.midgard.homeip.net>
In-Reply-To: <4A12225E.6000900@rawbw.com>
References:  <4A00B3FA.5050905@rawbw.com> <alpine.BSF.2.00.0905060818230.28367@wojtek.tensor.gdynia.pl> <4A11C53C.3090008@rawbw.com> <4ad871310905181337o6c368e14mccccf884f6839737@mail.gmail.com> <4A12225E.6000900@rawbw.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, May 18, 2009 at 08:07:10PM -0700, Yuri wrote:
> Glen Barber wrote:
> > No, that is not reasonable (but in a perfect world, would be correct).
> >
> > You are limited by the bus speed and controller buffer speed
> > (whichever is slower).
> >   
> 
> 
> My motherboard has PCI Express v2.0, that is 500MB/s limitation.

That's 500MB/s for each PCI-E 2.0 lane.  The PCI-E lanes originating at your
southbridge (Intel ICH10R) are only PCI-E 1.0 however for a max of 250MB/s 
per lane (in each direction.)  This is irrelevant however since the built-in
SATA controller is not attached to any PCI-E lanes.

What is relevant is the connection between the northbridge and southbridge.
In your case that connection is equivalent to a PCI-E x4 bus, for a max
speed of 1GB/s (in each direction.)

> Also it has HD/RAID controller Intel ICH10R Southbidge that has peak 
> SATA Data Rate 300MB/s.

That's 300MB/s for each SATA channel.  Unless you use a port-multiplier
(which is poorly supported in FreeBSD) you will only have a single disk
per SATA channel.

> So this brings me to 300MB/s limit.
> So 3X harddrives at 85MB/s will still scale. And 4X will already show 
> the limitation.

Actually the buses involved sets the limit at 1GB/s (the
Nortbridge-Southbridge connection.)  There are likely to be other limits
in play however such as the maximum throughput of the SATA controller,
or the max load the CPU can sustain, or the transfer speed to RAM.

Of these I suspect the controller itself will be the major bottleneck in
your case closely followed by the CPU.  Exactly how much they can support
is not so easily predicted however - it will have to be measured.


-- 
<Insert your favourite quote here.>
Erik Trulsson
ertr1013@student.uu.se



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20090519050630.GA16959>