Re: Experience with Raid (Good or Bad)
Date: 1996/04/10
Message-ID: <316C2C4D.5968_at_ray.nlm.nih.gov>#1/1
Robert Gerber wrote:
>
> Flash <cd001522_at_interramp.com> wrote:
> > I am in the market for a Raid solution to run a
> > variety of DBMS's on (Informix, Oracle, Sybase)
> > and have come down to a short list of four
> > vendors:
> > Sun SSA
>
One of the sites which I work at has between 100-200 Gigs of
Sybase disk space on Sun Sparc Storage Arrays- the customer
ends up using sybase to provide the mirroring, not the hardware
controller- All things considered, the SSA is still relatively
"young" in my view, and is continuously maturing- it was not a
full featured, high-avail. system when I was first introduced
to it 1+ years ago... I sort of believe purchasing an im-mature
RAID device is an oxymoron (sp?) one would choose RAID for
resilence- and traditionally, new technology is not known
for this- Sun is still adding new functionality / features to this
system---
Again- like a previoulsy posted reader replied- we almost always
have to take down the machine to replace a failed drive in an
SSA, unless the disk platters/mirrors are planned out perfectly-
>
> > Dec Storage Works
>
I work at another site, and have spent MANY hours configuring
RAID 1 and RAID 5 on the Dec Storage Works... it a nice system,
works pretty well... but is relatively hard to configure- get some
training if you plan on configuring the drives/raidsets/storageset
yourself- the commands are straight forward- its just that the docs
2 inches thick-
> > DG Clarion
>
I'm definitely biased here- we have over 1 terabyte on DG clarions-
about 20 clarions containing 400Gigs of Sybase databases, and another
23 clarions containing 1,000 Gigs+ of unix/image datasets-
they're slickthey'
re expensive- they're great- we use them with and with/out cache
always with 3 power supplies, always with 2 SP's (internal controllers)
We've put sybase raw partitions on RAID 1, RAID 5, and RAID 1/0 systems... We've used .5 Gig, 1 Gig, 2 Gig, and 4 Gig drives, We always use 2 SCCI cables/channels for extra bandwidth....
In 3+ years of operation, we've had only one critical failure when 2 power supplies broke at the same time... And amazingly (knock on wood) we have NEVER had to use any tape backups!!! Drives do fail, the system continues to run, we call up DG, they replace the broken drives the next day---
A long time ago, I calculated with over 500+ drives, at 300,000 hours MTBF, a drive will break at one of our sites every 2-3 weeks- RAID was the solution in our case-
Joel Plotkin Received on Wed Apr 10 1996 - 00:00:00 CEST