Date: Sun, 06 Jan 2002 17:59:20 -0500 From: Bill Moran <wmoran@potentialtech.com> To: "Kaven, Oliver" <Oliver_Kaven@ziffdavis.com> Cc: cdillon@wolves.k12.mo.us, cdillon@inter-linc.net, tlambert2@mindspring.com, advocacy@freebsd.org Subject: Re: Your freebsd-advocacy discussion Message-ID: <3C38D6C8.4050605@potentialtech.com> References: <B3F38D3937BA2646BBB30F0682E9F0270BF625@EDMSG007.media.ziffdavis.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Thank you for the clarification. I think, however, that you missed the point of my original post. That point was that FreeBSD often tends to outperform Linux under similar conditions, and I was curious to see if it would do so in the SMB arena. -Bill Kaven, Oliver wrote: > Gentlemen: > > I just recently stumbled across a discussion thread on freebsd-advocacy > relating to one of PC Magazine's past articles. I know that this thread > is fairly old, but would like to offer some insight into the testing > methodology if still needed. > > Here are some answers to the basic questions in the thread: > > Benchmark used: Netbench 7.0.2 > Client OS used : Win2K with most recent service pack at the time (SP2) > Client RAM used: RAM was a mixture of 256 and 512 MB > > I would like to stress that we did use Netbench as a performance > measurement tool. Netbench's number of clients does not relate to actual > users. One Netbench client generates traffic that corresponds to many > more clients in "real-life". > > - - - - - - - - - - > Concepts about NetBench's results > Here's a summary of some of the NetBench concepts you may want to keep > in mind as you look at your results: > > * NetBench's standard test suites are stress tests. > * Because one NetBench client generally stresses the server as > much as MANY actual users do, you can run test suites with a > relatively small number of clients and still get an accurate > measure of your server's performance. > * To get a valid measure of your server's performance, make > sure you reach a knee in the NetBench results curve. The > knee indicates that throughput is no longer increasing. (You > can check the results curve by plotting the server's total > throughput against the number of active clients.) > > Adding clients increases the total throughput ... up to a point. When > the overhead of managing the additional clients outweighs the advantage > of having more clients, throughput starts decreasing, causing a knee to > appear in the results curve. With NetBench you may see one of three > types of results curves: > > * A curve with a single knee that shows a very steep drop in > throughput as you add clients. > * A curve with a single knee that shows a gradual, sloping > drop in throughput as you add clients. > * A curve with a double knee. In this type of curve, the > throughput reaches its knee and then levels out for a while, > just as a plateau does, as you continue to add clients. > After you've added a certain number of clients, the > throughput drops off sharply. This type of curve often > indicates physical bottlenecks that affect the server's > performance. > > - - - - - - - - - - > > > I hope this helped a little and please, feel free to contact me directly > if you have any more questions. And yes, we do hope to include FreeBSD > into our testing in the future. Unfortunately , time restrictions often > force us to limit our testing to the platforms that relate most to our > readers environments and interests. > > > Have a healthy and Happy New Year! > > Oliver Kaven > ------------------------------------------------------- > Oliver Kaven > Project Leader, Network Infrastructure > PC Magazine Labs > 212-503-5283 > oliver_kaven@ziffdavis.com > ---------- > PC Magazine > 28 E 28th Street, 11th Floor > New York, NY 10016-7930 > 212-503-5100 > ---------- > > > > -- Bill Moran Potential Technology http://www.potentialtech.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-advocacy" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3C38D6C8.4050605>