In the past few months we have switched or production volume set from
JBOD to EMC Symmetrix 3700 with 16 x 23Gb spindles (each seen on the
3k as 6x4Gb) using 6 FWSCSI boards and 2Gb cache. We installed the
first configuration as RAID S. The (write) performance was so bad
that we had to reconfigure them RAID 1. The performance is still not
great and we see service times comparable to yours, i.e., 60ms on some
of the EMC drives. As you can see, the EMC will not resolve your IO
problems. Application design appears to be the main problem here and
we are now addressing that issue (although I'm sure that the limited
number of spindles does not help). Also watch out for the pricing.
EMC has priced later add-ons. i.e., disk and cache much higher than
the original configuration. We are now ordering 18Gb drives (will
configure them to be seen as 2 x 9Gb by the 3k) and adding another 2Gb
cache. For the system volume set we use an HP Model 20 array (raid
1). That appears to be working out fine.
-----Original Message-----
From: Lee Gunter [mailto:[log in to unmask]]
Sent: Tuesday, April 28, 1998 2:06 PM
To: [log in to unmask]
Subject: Re: Arrays on MPEXL_SYSTEM. . .
Hi, Denys,
We use one of each of the arrays in your list, in addition to a
C2258HA,
all of which totally constitute our system volume set. I have
noticed
performance degradation due, most likely, to the mismatched sizes of
the
three arrays and because of the low number of spindles. Disk service
times
on these three volumes can reach horrendously high values, IMO --
40-60ms
--, and utilization is practically 100% on all three during peak
usage
hours. Part of my plans for this year include migrating $SYS (my
short
name) to our EMC Symmetrix array, both for performance and redundancy
protection.
FWIW, these old Cascade-series arrays and all their other cousins in
our
data center have been extremely reliable in providing adequate RAID-3
protection. We never experienced a minute's downtime due to a
physical
disk failure. We do have a mech (HDA) fail from time to time, but HP
has
always been timely about getting us a replacement mech for a hot-swap
and
rebuild.
Lee Gunter
Regence Blue Cross Blue Shield of Oregon / Regence HMO Oregon
mailto:[log in to unmask]
voice...503-375-4498 fax.....503-375-4401
==========================================================
The opinions expressed, here, are mine and mine alone, and do not
necessarily reflect those of my employer.
From: [log in to unmask] on 04/28/98 09:56 AM
Please respond to [log in to unmask]
To: [log in to unmask]
cc: (bcc: Lee Gunter/BCBSO/TBG)
Subject: Arrays on MPEXL_SYSTEM. . .
X-no-Archive:yes
Gentle HP 3000 list members, I would like to know if anyone is using
the
C2254HA or the C2252HA, the disk arrays, as members of
MPEXL_SYSTEM_VOLUME_SET. If you do have these arrays on that volume
set,
how many do you have? Have you observed any system performance
degradation? Do you have any other comments?
Kind regards,
Denys. . .
Denys Beauchemin
HICOMP America, Inc.
(800) 323-8863 (281) 288-7438 Fax: (281) 355-6879
denys at hicomp.com www.hicomp.com
|
