Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!think.com!hsdndev!cmcl2!kramden.acf.nyu.edu!brnstnd
From: brnstnd@kramden.acf.nyu.edu (Dan Bernstein)
Newsgroups: comp.unix.admin
Subject: Can a remote side flood an Ethernet with under-200-byte TCP packets?
Summary: I very much doubt it, even with the latest routing technology
Message-ID: <7857:Apr1120:34:0791@kramden.acf.nyu.edu>
Date: 11 Apr 91 20:34:07 GMT
References: <11813@dog.ee.lbl.gov> <28077:Apr700:18:3191@kramden.acf.nyu.edu> <1991Apr9.145024.14464@murdoch.acc.Virginia.EDU>
Organization: IR
Lines: 20
X-Original-Subject: Re: Edit your subjects (was IRC Net Bandwidth)

In article <1991Apr9.145024.14464@murdoch.acc.Virginia.EDU> gl8f@astsun.astro.Virginia.EDU (Greg Lindahl) writes:
> In article <28077:Apr700:18:3191@kramden.acf.nyu.edu> brnstnd@kramden.acf.nyu.edu (Dan Bernstein) writes:
> >The big question: Can they still do it with 2-byte packets?
> The bigger question: why is IRC still in the subject when it seems to
> average 80 byte packets?

You're right; I've edited the subject line appropriately.

Chris reports that (relatively) new routers on the Internet run fast
enough that, in at least one case, a site achieved essentially full
Ethernet throughput via TCP over a wide-area link.

What percentage of this performance can it get with 2-byte packets?
20-byte packets? 200-byte packets, about the largest you'll ever see on
IRC?

After all, if IRC really is taking only 1.5% of total NSFNET resources,
I shouldn't be claiming that it takes twice as much.

---Dan