ADSM-L

Re: [ADSM-L] SATA disk?

2007-07-20 13:16:27
Subject: Re: [ADSM-L] SATA disk?
From: Wanda Prather <wprather AT JASI DOT COM>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Fri, 20 Jul 2007 11:48:24 -0500
Preach It Guys!

I have some customers with good SATA disk implementations that are very
happy.

I have some customers with some bad SATA disk implementations that are in
a very unhappy place.

It's all about understanding the performance characteristics of what you
buy, and matching it to your performance requirements.

Don't expect to haul a ton of concrete with a SmartCar.
And if it seems too cheap to be true.... there's probably a reason.

Wanda



> Allen is correct: lots of write back cache and you are good to go.
>
> The better solution IMHO is to use some fast SAS or FC disk as
> traditional disk device class pool fronting a bunch of SATA file device
> class.  You can then control the number of streams writing to the SATA
> disks and keep them in their happy place: serial I/O.
>
> We have probably 30 customers or so that are using various amounts (up
> to 200TB) of SATA disks in their STORServer/TSM environments.  Most, if
> not all, are very happy with their backup and restore performance.  For
> the larger sites, we're implementing some fast SAS or FC disk in front
> of SATA and improving performance by a bunch.
>
> My opinion: go forth and SATA baby!
>
>
> Kelly J. Lipp
> VP Manufacturing & CTO
> STORServer, Inc.
> 485-B Elkton Drive
> Colorado Springs, CO 80907
> 719-266-8777
> lipp AT storserver DOT com
>
> -----Original Message-----
> From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf 
> Of
> Allen S. Rout
> Sent: Wednesday, July 11, 2007 12:55 PM
> To: ADSM-L AT VM.MARIST DOT EDU
> Subject: Re: [ADSM-L] SATA disk?
>
>>> On Wed, 11 Jul 2007 13:09:31 -0500, Johnny Lea <jlea AT DIS.UMSMED DOT EDU>
> said:
>
>
>> I thinking the speed (7200 RPM) may be a problem.
>
>
> Good cache obviates that.  SATA is good for serial workloads, so if
> you're sending few (maybe <10) streams to it simultaneously, you should
> be good.
>
> It'll suffer faster under heavy random access contention than will e.g.
> FC disk.
>
>
> - Allen S. Rout
>

<Prev in Thread] Current Thread [Next in Thread>