Re: FreeBSD 5.3b7and poor ata performance

From: Scott Long <scottl_at_freebsd.org>
Date: Mon, 25 Oct 2004 14:45:37 -0600
Charles Swiger wrote:
> On Oct 25, 2004, at 3:49 PM, Scott Long wrote:
> [ ... ]
> 
>>> Your position is certainly reasonable: if a storage system is not 
>>> reliable, how fast it performs is something of a moot point.  :-)  
>>> However, this being said, a RAID-0 implementation needs to improve 
>>> performance compared with using a bare drive if it is to be useful.
>>
>>
>> Well, RAID-0 is a special case =-)
> 
> 
> Sort of, yeah.  It's hard to make generalizations about RAID performance 
> without considering each mode as a separate case...in which case, your 
> generalizations aren't very general.  8-)
> 
>> That said, putting discrete RAID
>> classes into the GEOM layer is something of a new adventure, so I'm
>> not surprised to hear about performance problems, even in RAID-0.
>> There might be extra data copies or path latencies that weren't planned
>> for or expected.  It's definitely something to look at.  But it's also
>> a very new subsystem, so it would be unfair to judge FreeBSD performance
>> with it.
> 
> 
> Oh, I'm not trying to throw stones your way, or at GEOM, or anywhere else.
> 
> By and large, you would be right to claim that RAID generally performs 
> less well than direct access to bare drives.  This conclusion is driven 
> as much by how frequently RAID-5 gets used compared with the less-common 
> RAID modes as anything else, however.  Someone who uses RAID-0 or 
> RAID-1,0 modes really does expect to see a performance improvement.
> 

RAID-0 yes, RAID-10 no, at least not for software RAID.  The machine 
winds up having to transfer the same data twice across the PCI bus,
twice through the controller, etc.  If the controller is on a simple
PCI-32/33 bus then it will quickly become saturated.

Anyways, having spent a good part of my career with RAID, I find that
I only use RAID-0 when I want to test system bandwidth, not when I
want to store data.  YMMV =-)

Scott
Received on Mon Oct 25 2004 - 18:47:23 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:19 UTC