HP3000-L Archives

August 2001, Week 2

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
"Costantino, Rocky" <[log in to unmask]>
Reply To:
Costantino, Rocky
Date:
Thu, 9 Aug 2001 11:57:09 -0400
Content-Type:
text/plain
Parts/Attachments:
text/plain (75 lines)
Thomas,

The physical drives will be carved into logical devices referred to as
hypervolumes. I believe the latest drive size offerings are 36GB and 73GB.
Since MPE does thrive on multiple spindles, 36GB drives would be preferred.

This is similar to a recent implementation that I helped design. The 36GB
drives were split 4:1, yielding 8.8GB hypervolumes (the reason that the
volumes are not 9GB is due to the fact that come space is reserved for
'gatekeepers', internal management structures for the Symm). The HP3000 will
see the HV's as physical drives which will become LDEVs. Use RAID-1
configuration!!!

NOTE: Use care in assigning devices to volume sets, as you want to avoid
spindle contention. The 4:1 splits will result in 2 primary devices (M1) and
2 mirrored devices (M2). This is for the RAID-1 config. The M1's should be
assigned to try to use one from each physical device per volume set.

I would use target-mode address, which is what you are most likely familiar
with, i.e., 0.6.0, 0.5.0, 0.4.0 versus lun addressing, i.e., 0.6.0, 0.6.1,
0.6.2. Spread the devices over as many fast-wide interfaces as you have
available. There is an ID available for Symmetrix devices for configuration
in SYSGEN (see IODFAULT.PUB.SYS).

PowerPath is not available for MPE. However, HP has High Availability Fail
Over (HAFO). Does not provide load balancing but does provide failover. I am
not sure what the official HP stance is on support for HAFO on EMC. It will
work, but may not yet be officially blessed by HP.

Hope this helps. Obviously each situation is different, so variations to the
above may be necessary.

Regards,

Rocky J. Costantino
Vice President of Professional Services
Computer Design & Integration, LLC
696 Route 46 West, Teterboro, NJ 07608
Phone: 201-931-1420 x224
Fax:     201-931-0101
[log in to unmask]
www.cdillc.com


-----Original Message-----
From: Thomas M. Root [mailto:[log in to unmask]]
Sent: Wednesday, August 08, 2001 3:38 PM
Subject: Configuration suggestions for a Summetrix storage array


Desert Schools Credit Union is about to install an EMC Summetrix storage
system and I would like to get some configuration recommendations from those
of you who have been down this path.

Summit, our software provider, recommends spreading our databases over many
disk drives to maximize the potential for I/O concurrency.  Even though the
logical disks configured on the 3000 don't directly correspond to spindles
on the Summetrix, I speculate that this is still good practice so that the
3000 can maximize the number of requests passed to the storage array.  Is
this correct?

As I understand it, each FWD SCSI controller can be configured with up to 15
target IDs and each ID may be configured with up to 16 logical devices.  Is
it better to use a few SCSI IDs with lots of logical devices, or use lots of
IDs with fewer devices?

Any other insights would be greatly appreciated.

Thomas M. Root <[log in to unmask]>
Desert Schools Federal Credit Union
Phoenix, AZ

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *

ATOM RSS1 RSS2