Path: news.easynews.com!easynews!feedwest.aleron.net!aleron.net!sfo2-feed1.news.algx.net!allegiance!logbridge.uoregon.edu!pln-w!spln!dex!extra.newsguy.com!newsp.newsguy.com!news.usenetgateway.com From: roobaron Newsgroups: comp.databases.oracle.server Subject: Re: Best things to look at when benchmarking Windows vs. UNIX Date: 9 Aug 2002 01:20:22 GMT Organization: dBforums Lines: 22 Sender: roobaron Message-ID: <1679938.1028854671@dbforums.com> References: <5003a2b9.0208071327.7f554c25@posting.google.com> NNTP-Posting-Host: p-915.newsdawg.com User-Agent: forums@dbforums.com X-Original-NNTP-Posting-Host: ns.spikeball.com X-Original-Trace: 9 Aug 2002 01:20:22 GMT, ns.spikeball.com Xref: easynews comp.databases.oracle.server:157319 X-Received-Date: Thu, 08 Aug 2002 18:47:39 MST (news.easynews.com) Benchmark the app on what the users require. if they require response time, benchmark that if they require throughput, benchmark that if they (company in this case) requires high numbers of concurrent users... benchmark that if there is a combo, you need to benchmark individually and combos i.e. how does number of users affect throughput or response time. Then push the envelope to see when the database/OS/app breaks, this will help determine scalarability. The best way to display the results is graphically if you can, use some basic stats (regression) for colrelation between response time, throughput and concurrent users. Have Fun -- Posted via dBforums\nhttp://dbforums.com