Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!wuarchive!rice!rice!sun-spots-request
From: fredc@umrisca.isc.umr.edu
Newsgroups: comp.sys.sun
Subject: Re: Price of comparable high-end PC vs. Sun SLC
Keywords: Hardware
Message-ID: <1990Oct7.221451.25527@rice.edu>
Date: 7 Oct 90 21:30:00 GMT
Sender: sun-spots-request@rice.edu
Organization: Sun-Spots
Lines: 25
Approved: Sun-Spots@rice.edu
Originator: spots@walhalla.rice.edu
X-Sun-Spots-Digest: Volume 9, Issue 328, message 4
X-Refs:  Original: v9n302, Replies: v9n307

In article <1990Sep4.232304.16861@rice.edu> oliveau@cs.ucla.edu
(Greg Oliveau) writes:
>Apparently you can hook up around 20 SLC's to a single disk (or SLC
>server) and not see significant performance degredation depending on what
>you do - this of course was from a Sun salesman!

I saw a system set up like this (DACNet at the Design Automation
Conference), and it was trash!  The reason: when you are running Open
Windows, an 8M SLC will thrash memory.  Put twenty of those running swap
off the same disk and you have a disaster.  If you run any windowing
system at all, anything with a local disk for swap will outperform a
diskless node with the same amount of real memory, unless that amount
happens to be large enough to avoid swapping altogether.  I have found
that I tend to use up whatever real memory is there by enhancing my basic
layout, so I'll never have enough real memory to do that :-)

An aside: Sun was there in force at DAC, knew about the mess called
DACNet, yet they made no attempts to remedy the problem.  I thought that
such negligence shows a remarkable arrogance, not unlike another 3-letter
company we all love to hate.

Fred Clauss                  INTERNET:  fredc@isc.umr.edu (preferred)
Intelligent Systems Center    		or fredc@umree.ee.umr.edu
University of Missouri        UUNET:	{occrsh|sunarch}!umree!fredc
Rolla, MO 65401		     BITNET:    S081192@UMRVMA