Path: utzoo!utgpu!cunews!bnrgate!brchh104!brchs1!bnr.ca!rice.edu!sun-spots-request
From: chris@com50.c2s.mn.org (Chris Johnson)
Newsgroups: comp.sys.sun
Subject: Re: SPARCstation file system read/write bandwidth???
Keywords: No Digest Subjects during Flush
Message-ID: <4014@brchh104.bnr.ca>
Date: 27 Jun 91 20:22:00 GMT
Sender: news@brchh104.bnr.ca
Organization: Sunspots, Flush Mode
Lines: 27
Approved: sun-spots@rice.edu
X-Original-Date:    Wed, 12 Jun 1991 17:44:46 -0500

In article <3767@brchh104.bnr.ca> jes@wookie.ess.harris.com (Jim Stroud) writes:
>Does anyone have any feeling ( or relevant data ) regarding the average
>sustained speed of a SPARCstation file system.
>We are using a SPARC to accumulate data in a random I/O file and access
>previously captured data in another part of the file.
>What is the average sustained READ/WRITE bandwidth for this type of
>operation???

I'd suggest running your own benchmarks, for starters.  There are quite a few
variables involved, and without specifying a few more things, there's no way
anybody can give you anything relevant.

For example:  Which model SPARCstation or SPARCserver are you running?
Which level of the SunOS are you running (Sun claims to have significantly
speeded up the filesystem in 4.1.1)?  What type of disks are you using?
What are the process characteristics?  Is your system CPU or IO bound or
neither?

We can give you some general ideas about bandwidth upper limits if we know
the answers to the above questions.  Some people might have enough data and
experience with an application similar to yours to even give you some more
applicable numbers, if they can pin down some of the variables.

-- 
   ...Chris Johnson          chris@c2s.mn.org   ..uunet!bungia!com50!chris
 Com Squared Systems, Inc.   St. Paul, MN USA   +1 612 452 9522