[nycbug-talk] Fwd: Adaptec AAC raid support

Okan Demirmen okan
Mon Mar 21 18:04:12 EST 2005


On Mon 2005.03.21 at 18:01 -0500, bruno wrote:
> On Mon, Mar 21, 2005 at 05:09:07PM -0500, Okan Demirmen wrote:
> > On Mon 2005.03.21 at 16:57 -0500, bruno wrote:
> > > On Mon, Mar 21, 2005 at 03:57:28PM -0500, Marco Scoffier wrote:
> > > > On Mon, Mar 21, 2005 at 03:31:10PM -0500, bruno wrote:
> > > > >If you do need performance software raid is sometimes not enough.
> > > > >
> > > > I was wondering if anyone had numbers on this.  Are we talking +5% cpu
> > > > utilisation and -5% throughput or +20% and -20% or more ??  Searched
> > > > a bit for 
> > > > numbers but came up dry.
> > > 
> > > I don't use numbers and I'm afraid I have none to offer. :) Those are
> > > more for managers. I've seen visible differences between hardware and
> > > software RAID, on both PC-like servers and big SUNs.
> > 
> > imho, numbers are not just for managers and the ones putting down
> > the cash. numbers are important, as you eluded, to the application
> > and environment. i think numbers can make or break you when it comes
> > to scaling and all the things one needs to think of whilst researching
> > the solution for your environment. not picking on you, but "visible"
> > and "feel" are words i don't like hearing in the context of
> > hardware/software performance. numbers are key in my book.
> 
> Sure. All I gave was my opinion, which you can take for what it is, my
> opinion. I'm glad you educated everyone not to base their purchase on
> "visible" and "feel" words, that sure was not what I wanted. Perhaps
> my post was then useless. I only hoped to help someone. Oh well.

no. you're post was not useless. i didn't intend to knock on your
post (re: not picking on you). i merely wanted to make a point about
something i've seen more and more lately, hence my $.02...that's
all ;)

okan

-- 
Okan Demirmen <okan at demirmen.com>
PGP-Key: http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xB3670934
PGP-Fingerprint: 226D B4AE 78A9 7F4E CD2B 1B44 C281 AF18 B367 0934




More information about the talk mailing list