*BSD News Article 59931


Return to BSD News archive

Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!news.mel.connect.com.au!munnari.OZ.AU!news.ecn.uoknor.edu!paladin.american.edu!europa.chnt.gtegsc.com!gatech!newsfeed.internetmci.com!vixen.cso.uiuc.edu!sdd.hp.com!hamblin.math.byu.edu!park.uvsc.edu!usenet
From: Terry Lambert <terry@lambert.org>
Newsgroups: comp.unix.bsd.netbsd.misc,comp.unix.bsd.bsdi.misc,comp.unix.solaris,comp.unix.aix
Subject: Re: ISP hardware/software choices (performance comparison)
Date: 18 Jan 1996 04:22:31 GMT
Organization: Utah Valley State College, Orem, Utah
Lines: 40
Distribution: inet
Message-ID: <4dkhu7$27e@park.uvsc.edu>
References: <4cmopu$d35@vixen.cso.uiuc.edu> <4dic25$f15@maureen.teleport.com>
NNTP-Posting-Host: hecate.artisoft.com
Xref: euryale.cc.adfa.oz.au comp.unix.bsd.netbsd.misc:2024 comp.unix.bsd.bsdi.misc:2182 comp.unix.solaris:57597 comp.unix.aix:69012

connect@transport.com (Net Connection) wrote:
] while the SPEC benchmarks (and several others) indicate the Pentium
] systems to be just as fast (even faster) than similar made for Unix
] boxes, one parameter is not often talked about - Socket Bandwidth.
] 
] LMbench tests for socket bandwidth, and the Sparc - Dec - HP etc.
] machines ring in at 25 to 100 times faster than the Pentiums on that
] one parameter. 
] 
] I don't know yet exactly how important that measure is, but it seems
] to me that it could be very important in Web Service applications.
] 
] I would appreciate any input on this - I'm sure others would also


I would consider it *very* important.  I/O bandwidth used to
be one of the overriding factors in favor of SPARC hardware;
for newer SPARC hardware (which is simply not cost-competitive
with the commodity Intel hardware), it still is.

Whether it would be worth it for an ISP really depends on
whether the host machine or your NSP pipe size is where your
I/O bandwidth is being limited.


I would caution against taking any benchmark result as a
"single figure of merit" when choosing an ISP platform,
however.

LMBench has at least one test whose merits have been hotly
debated several times in various groups on various lists
(not socket bandwidth) as preferring "dangerous" async
metadata operations in the host FS implementation.


                                        Terry Lambert
                                        terry@cs.weber.edu
---
Any opinions in this posting are my own and not those of my present
or previous employers.