Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!gem.mps.ohio-state.edu!uakari.primate.wisc.edu!dogie.macc.wisc.edu!uwvax!umn-d-ub!umn-cs!herndon
From: herndon@umn-cs.CS.UMN.EDU (Robert Herndon)
Newsgroups: comp.arch
Subject: Re: Workstation speed comparisons
Summary: addendum to Kern's law of _multiuser_ computer performance
Message-ID: <16869@umn-cs.CS.UMN.EDU>
Date: 9 Nov 89 01:16:40 GMT
References: <12850@boulder.Colorado.EDU> <1989Oct19.153249.25491@cs.rochester.edu> <313@voa3.UUCP>
Distribution: na
Organization: University of Minnesota, Minneapolis
Lines: 32

In article <313@voa3.UUCP>, ck@voa3.UUCP (Chris Kern) writes:
> In article <1989Nov5.182954.27464@world.std.com> bzs@world.std.com
> >SHEIN'S LAW OF COMPUTER PERFORMANCE:
> >	The speed of a computer is measured from the time
> >	at which a question is asked to the time at which
> >	the question is answered. All other performance
> >	benchmarks are lies.
> Kern's law of _multiuser_ computer performance:
> 	All multiuser computers take the same amount of time
> 	to execute your program.  They just support different-sized
> 	user populations.
> Chris Kern			     Voice of America, Washington, D.C.

Herndon's addendum to Kern's law of _multiuser_ computer performance:

	Kern's law applies where the software environment is
	fixed.  If software is a variable, the more convenient
	the software is, the slower the hardware is at running
	your application.

Corollary:
	The response time of an open multiuser system is inversely
	proportional to the convenience of the environment.

 This follows from the observation that the easier it is to do
something useful on a system, the more people will do on it, until
the system is so busy that they can better spend their time elsewhere.
-- 
Robert Herndon				Dept. of Computer Science,
...!ihnp4!umn-cs!herndon		Univ. of Minnesota,
herndon@umn-cs.ARPA			Minneapolis, MN  55455
herndon.umn-cs@csnet-relay.ARPA