RE: DASD arm worksheets

IBM has changed to a I/O per second calculation for sizing disk on the newer 
models. The way I do it is to collect performance data to determine the current 
average I/Os per second. I then add a growth factor and multiply that by the 
number of drives on the system. I then divide that number by the number of I/Os 
per second the new disk subsystem can support per drive. That can be anywhere 
from 40 to 60 depending on how conservative you want to be. This will give an 
estimate of the number of drives to propose.

Kendall Kinnear
Consulting System i5 Architect
Stonebridge

Phone: 214-676-3146
email:   kendall.kinnear@xxxxxxxx

Sent from Windows Mobile 5

-----Original Message-----
From: "ChadB@xxxxxxxxxxxxxxxxxxxx" <ChadB@xxxxxxxxxxxxxxxxxxxx>
To: "Midrange Systems Technical Discussion" <midrange-l@xxxxxxxxxxxx>
Sent: 5/2/06 10:13 AM
Subject: DASD arm worksheets


Does anyone know if IBM still publishes the "DASD Arms Required for
Performance" type worksheets for the i5 models?  I've used them in the past
as an additional measure of insurance when designing a new system, but the
last time I used one was for an 800.  We're starting to spec out a new 520
to replace the 800 and with the Domino and Websphere workloads I KNOW we've
had disk arm/controller issues in the past, so I want to ensure a good
config.  The 800 was originally spec'ed by the previous admin/BP with a
5702 controlling 6 mirrored drives and the box performed horribly.  I got
it switched over to a RAIDed 2757 and made a positive impact (and want that
to carry over to the new box!).

In addition to the DASD arms worksheet, i'm looking for some input on the
757mb cache and 40mb cache RAID controllers available for the i5s.  We're
hoping to keep this box to just a system enclosure and I think with the
520s all the 757mb controllers are not permitted in the system enclosure.
Is this true?  If so, i'm wondering how a 40mb controller will fare in
comparison to the current 2757 (757mb cache) we're running.  I would hate
to take a step backwards on the disk cache... Domino seems to love the
increased disk cache.