HP3000-L Archives

September 1998, 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:
Neal Kazmi <[log in to unmask]>
Reply To:
Neal Kazmi <[log in to unmask]>
Date:
Thu, 10 Sep 1998 15:26:28 -0700
Content-Type:
text/plain
Parts/Attachments:
text/plain (36 lines)
<SNIP>
>Does having 6 4g drives produce more IO/sec  than 3 9g drives?
</SNIP>

Many years ago...

I accepted that it would be a good idea to switch from five pairs of
mirrored 670MB drives (3.3GB) to a single 5.4 HA disc array.  The primary
application consisted of many processes most of which did Image calls.
Reports started taking three to ten times as long.  :(

It took switching to two 2.7GB arrays to bring the run times back within
reason.

MPE was doing a very good job of scheduling processes for when data is
available from many drives.  Even though there are several spindles in the
disc array, it is the number of spindles that MPE can queue on that matters.

Lesson learned...

If you have a large number of independent processes, more small drives are
preferred to fewer large drives.

Digression...

I extrapolated this to the point where it should be a good idea to a very
large number use those un-marketed "Kitty Hawk" drives (20-40MB).  If only
there was enough back plane for the interface cards.

neal

Neal Kazmi ([log in to unmask])
Voice:(800)682-0200
FAX:(360)568-2923
<http://www.minisoft.com> (Home Page for MiniSoft, Inc.)

ATOM RSS1 RSS2