Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!news.rmit.EDU.AU!news.unimelb.EDU.AU!munnari.OZ.AU!news.ecn.uoknor.edu!qns3.qns.com!imci4!newsfeed.internetmci.com!news.texas.net!news1.best.com!nntp1.best.com!shellx.best.com!not-for-mail From: rcarter@shellx.best.com (Russell Carter) Newsgroups: comp.unix.bsd.freebsd.misc Subject: Re: FreeBSD as a router Date: 29 Apr 1996 13:45:52 -0700 Organization: Best Internet Communications Lines: 104 Message-ID: <4m39q0$bqu@shellx.best.com> References: <4lfm8j$kn3@nuscc.nus.sg> <4m0mht$qcl@skipper.netrail.net> <4m1pjc$nle@itchy.serv.net> <DqMLuC.2xo@ritz.mordor.com> NNTP-Posting-Host: shellx.best.com In article <DqMLuC.2xo@ritz.mordor.com>, Chris Mauritz <ritz@ritz.mordor.com> wrote: >Sean T. Lamont (zeno@serv.net) wrote: >: In article <4m0mht$qcl@skipper.netrail.net>, >: Nathan Stratton <nathan@netrail.net> wrote: >: >Steinar Haug (sthaug@nethelp.no) wrote: >: >: >: >: You may want to check out SDL Communications and their T3 card: >: > >: >You want to write the FreeBSD driver? >: > >: >: >: for the price of a cisco 7010 + HSSI, I'd certainly give it a shot! :) >: (Not there yet, but looking forward.) >: >: By the way, whoever it was that was indicating some fundamental flaw >: in unix that you couldn't do over 2M is highly confused ; higher-end >: unix boxes such as RS6K's have been doing this for years. > >I agree. Perhaps this was true of slower machines, but I'm willing >to bet that any modern pc with a good motherboard and fast cpu can >do much better than 2M/sec. In terms of cost, the pc enjoys a huge >advantage...you can build a 64M P166 box for roughly $2k these days. > >Regards, > >Chris >-- >Christopher Mauritz | For info on internet access: >ritz@mordor.com | finger/mail info@ritz.mordor.com OR >Mordor International | http://www.mordor.com/ >201/212/718 internet access | Modem: (201)433-7343,(212)843-3451 Hmm... I know I tested routing a couple of months ago, aha! Here is what I measured with the latest version of netperf, systems use DC21140 NICs, (two in router), one pair direct connected and one pair connected through a Trendnet hub. Pipelined burst SRAM (essential!) P5-133/166/133 CPUs. All with 32 MB. FreeBSD 2.1-STABLE. I haven't posted these before because I was a little disappointed, but hell they are just a shade over 20 times faster than ah has been claimed is the max for a PC Unix router. TCP REQUEST/RESPONSE TEST to gelifast Local /Remote Socket Size Request Resp. Elapsed Trans. Send Recv Size Size Time Rate bytes Bytes bytes bytes secs. per sec 16384 16384 1 1 60.00 2056.66 9216 41600 1 1 60.01 2454.44 9216 41600 516 4 60.01 1878.81 16384 16384 1 1 60.01 2031.38 16384 16384 1 1 60.00 2013.41 16384 16384 1 1 60.00 2030.44 16384 16384 1 1 60.00 2038.81 TCP STREAM TEST to gelifast Recv Send Send Socket Socket Message Elapsed Size Size Size Time Throughput bytes bytes bytes secs. 10^6bits/sec 32768 32768 1 60.01 0.32 32768 32768 4 60.01 1.25 32768 32768 16 60.00 4.20 32768 32768 64 60.00 10.76 32768 32768 256 60.00 21.77 32768 32768 1024 60.00 32.03 32768 32768 4096 60.00 38.32 32768 32768 16384 60.00 41.09 32768 32768 65536 60.01 39.78 57344 57344 4096 60.01 39.33 57344 57344 8192 60.00 40.11 57344 57344 32768 60.04 41.09 32768 32768 4096 60.01 39.56 32768 32768 8192 60.01 40.99 32768 32768 32768 60.01 40.49 8192 8192 4096 60.01 39.99 8192 8192 8192 60.01 40.98 8192 8192 32768 60.01 40.27 UDP UNIDIRECTIONAL SEND TEST to gelifast Socket Message Elapsed Messages Size Size Time Okay Errors Throughput bytes bytes secs # # 10^6bits/sec 32768 64 60.00 1088107 0 9.28 32768 60.00 391019 3.34 32768 1024 60.00 580151 0 79.21 32768 60.00 381200 52.04 32768 1472 60.00 483611 0 94.92 32768 60.00 311178 61.07 Looks pretty good to me... BTW, unrouted TCP throughput peaks at about 83 Mb/s over 100BASE-TX. Later, Russell