Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 5 Oct 1998 16:00:28 -0700
From:      perry@zso.dec.com (Reginald Perry)
To:        "'Eivind Eklund'" <eivind@yes.no>, <freebsd-chat@FreeBSD.ORG>
Subject:   RE: PC Magazine 10/20/1998 Article about FreeBSD
Message-ID:  <69CAF7F9AF57D2118D9A0000F881B4DD02F304@zsoexc1.zso.dec.com>
In-Reply-To: <69CAF7F9AF57D2118D9A0000F881B4DD06BB49@zsoexc1.zso.dec.com>

next in thread | previous in thread | raw e-mail | index | archive | help
DUH! Well of course NT was able to increase from ~200 requests/second to
catch up with FreeBSD. The FreeBSD box had saturated the network and had
nowhere else to go. The NT box still had 400 requests/second worth of head
room before network saturation! So who is going to write the author and
request a correction?

-----Original Message-----
From: Eivind Eklund [mailto:eivind@yes.no]
Sent: Monday, October 05, 1998 3:13 PM
To: Reginald Perry; freebsd-chat@FreeBSD.ORG
Subject: Re: PC Magazine 10/20/1998 Article about FreeBSD


On Mon, Oct 05, 1998 at 01:53:35PM -0700, Reginald Perry wrote:
> They didn't say. This is the relevant paragraph:
>
> "We tested FreeBSD in one of its most common applications: Web serving. We
> set up two Dell PowerEdge 2200 servers with 128MB RAM and a single Pentium
> II CPU, installing FreeBSD with Apache 1.3.0 on one and Windows NT 4.0
with
> IIS 4.0 on the other. On our ZD WebBench 2.0 tests, performance leveled
off
> quickly; memory was the bottleneck for both NOSs. FreeBSD outperformed
> Windows NT by a sizable margin, however, as you increase RAM, Windows NT
> surpasses FreeBSD because of a cache limitation in Apache and FreeBSD."
>
> At the bottom of the page, is a WebBench graph of clients on the X axis
and
> requests/second on the Y axis that shows both leveling off at about 8
> clients with NT starting to level off above 4 clients and FreeBSD leveling
> off very sharply at 8 clients. The level is at ~200 requests/second for NT
> and ~600 requests/second for FreeBSD, if I am extrapolating this graph
> correctly. The graph measures out to 60 clients. Of course they failed to
> show a graph for average maximum requests/second vs. amount of RAM.

So FreeBSD has about 3x higher performance than NT on the same hardware?

Anyway; 600 requests/second is about 10MBit/(600*8) = 2184 bytes
transfer per request.  This fit pretty well with the fact that they're
trying to emulate a typical web-server load (according to the WebBench
description at http://www.zdnet.com/zdbop/webbench/1main/1wrktree.htm).

If that is IT (ie, they're using a 10MBit NIC) I'm not surprised at
the sharp cutoff - I'd expect a sharp cut-off around the capacity of
the network :-)

Eivind.


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-chat" in the body of the message



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?69CAF7F9AF57D2118D9A0000F881B4DD02F304>