Re: asm on san

From: gym dot scuba dot kennedy at gmail <kennedyii_at_verizon.net>
Date: Mon, 15 Dec 2008 21:45:27 GMT
Message-ID: <XdA1l.1806$c35.827@nwrddc02.gnilink.net>

"Robert Klemme" <shortcutter_at_googlemail.com> wrote in message news:6qo0nrFdqgb7U2_at_mid.individual.net...
> On 14.12.2008 21:47, gym dot scuba dot kennedy at gmail wrote:
>
>> THe whole thing is a chain and one must understand the chain to identify
>> bottlenecks. For example, you could have data stripped across 1,000
>> disks and only have 1 CPU on the san, Your limitation is the CPU in that
>> example.
>
> Not necessarily: you can cripple performance even more by bad application
> design. :-)
>
> Cheers
>
> robert

Robert,
We are in agreement. Poor application design is a major cause of in the toilet performance. Absolutely. The point I meant to make was that just having a lot of disks stripped allows the disks to serve up a ton of bandwidth, but if you don't have sufficient CPU or IO channels etc. then it is meaningless.

But yes, it is paramount to start at the application level. Proper application design allows you to run well on a much wider range of hardware. Yes, that is vital, agreed.
Jim Received on Mon Dec 15 2008 - 15:45:27 CST

Original text of this message