Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!purdue!gatech!hubcap!Hemant
From: hemant@ms.uky.edu (Hemant Rotithor)
Newsgroups: comp.parallel
Subject: Question on Load Balancing
Message-ID: <5944@hubcap.clemson.edu>
Date: 7 Jul 89 13:08:49 GMT
Sender: fpst@hubcap.clemson.edu
Lines: 25
Approved: parallel@hubcap.clemson.edu

A Question About Load Balancing

I wish to test the performance of a load balancing scheme on
an actual network of VaxStations running Ultrix 3.0.

I have the following questions:

1. What are good performance metrics to evaluate the performance
of load balancing, please suggest some good references.

2. Mean Response Time has been used by many researchers as a metric
to evaluate the performance of load balancing.
If the performance is evaluated on an actual distributed system by
subjecting each node to a script of commands, then should the "response time"
consist of service times of all the commands or should it include
the "wait times" of the commands also?
The papers I have read on this topic do not explain this clearly and 
therefore the confusion on my part.

I will appreciate any responses, thanks in advance.

-- 
Hemant G. Rotithor (606)258-2656       * E-mail: hemant@ms.uky.edu or
Elect. Engg., 206 Annex.               *         hemant@engr.uky.edu     
University of Kentucky, KY 40506-0046