NetNews Usenet Archive 1992 #16/NN_1992_16.iso/spool/comp/dcom/cellrel/239

Text File | 1992-07-23 | 1.8 KB | 39 lines

Newsgroups: comp.dcom.cell-relay
Path: sparky!uunet!cs.utexas.edu!sdd.hp.com!usc!zaphod.mps.ohio-state.edu!menudo.uh.edu!casc.math.uh.edu!stan
From: [e-mail] (Stan Hanks)
Subject: Re: Question
Message-ID: <[e-mail]>
Followup-To: <[e-mail]>
Sender: [e-mail] (USENET News System)
Nntp-Posting-Host: casc.math.uh.edu
Organization: Center for Advanced Scientific Computation, Houston Tx.
Date: Fri, 24 Jul 1992 07:14:25 GMT
Lines: 26

In article <[e-mail]> you write:
>>My understanding is that cell relay, in particular ATM, has really poor
>>efficiency (~40%) when used in a LANish mode, and that under 150 Mb/s
>>or so, it's hardly worthwhile to implement.
>
>....  The efficiency is then 44/53 =83%.   This
>is result is independent of any particular physical considerations.


Reference the Berkeley papers from TENET project. I think the actual
title is "Efficienct of TCP/P over ATM" or some such. It used to be
up for FTP somewhere at tenet.berkeley.edu.  There's *ALWAYS* more
to life than simple calculations. Think about what sending say a 50
byte (total length) packet does to your efficiency if you have a fixed
44 byte per cell payload: one cell full, one cell with 6 bytes of data
which makes it only 11% used. A little of that goes a long way to hose
up the overall efficiency of the network.

On the 150 Mbps comment, I can't tell you the source. But it's from someone
who does chip design for ATM and related technologies.

--
Stanley P. Hanks      Principal Scientist,      Technology Transfer Associates
U.S. Snail:        P.O. Box 2087,   Bellaire    TX    77402-2087
FedEx/Shipping:    5303-F Glenmont, Houston     TX    77081
e-mail: [e-mail]   voice: [phone]     fax: [phone]