On Fri, 19 May 2000, Ard van Breemen wrote:
> On Wed, May 17, 2000 at 07:19:43PM -0300, Christian Robottom Reis wrote:
> > If you want to know what slow means, I'll post some SW-Raid readbalanced
> > RAID1 benchmarks *grin* Mika rules!
> (*) Yes! I want to know what slow means!
> I get some very nice benchmarks, and .sig for free!
> ( ) No! I do not want to know what slow means!
> I do not get the benchmarks, but I do get a .sig for free!
Right choice. About slow; well, I meant _Hardware RAID1 is slow_; if you
see the Software RAID1 performance I've obtained on very similar hardware,
you'd understand that it makes HW Raid seem like a non-option. Only when
you take into account the dreaded hotswap and cpu usage can one float back
into a more competitive comparison.
> I'm trying to collect some data on [HS]w-raid.
> But somehow these seem to be always incomplete.
> If someone posts raid-results with bonnie/tiobench, please post
> the following with that:
Bonnie doesn't measure the real issue with Raid1 - it doesn't relate much
to serving as a matter of fact, since all servers run many programs
simultaneously (that's what serving means IMO). I do have some bonnies.
> Disks used (details please, not some 10k something).
> Controller used.
> Drivers used (kernel revision and driver revision when not vanilla, and all
> patches of course).
> Logical drive setup. We want to know at what point at the disk the test
> takes place. I love to see tests done on emtpy disks, so that tests
> will take place somewhere around the edges of the disks.
> Raid setup of course...
Have to dig stuff up and see if I have this all. Could very well be this
is gone forever, but then again we can always benchmark some more.
> Globally speaking, the results are the most important, but testing raid f.i.
> with drives that can reach the absolute max of (hold your breath) 1Mb/s
> seems not very sane.
Yeah, I know. Most of this was decent hardware, but to prepare a 100%
complete benchmark framework takes resources and time. I'll have this
ready when the burn-in lab is done, but since we're still putting this all
together.. [yawn]. Let me see what I can find.
Cheers,
--
/\ Christian Reis exists solely to answer [EMAIL PROTECTED]
\'`/ Async Free SW Development | http://async.com.br | +55 16 274 2497
