Aucbvax.6477
fa.works
utcsrgv!utzoo!decvax!ucbvax!works
Sun Mar 14 23:57:58 1982
Re: Ethernet Doomed?
>From mo@Lbl-Unix Tue Mar 9 05:41:43 1982
"Rumors of my recent demise have been greatly over-exagerated."
-Mark Twain
The market survey which pronounced the death of Ethernet is not
completely accurate and grossly oversold, but could you think
of a better scam to sell zillions of your document?
All the same, the comments by Xerox about broadband are equally
ill-informed. I have spent quite a while comparing the two
schemes and have come to the following conclusions.
1) Broadband does require a bit more initial planning because
you are wiring the world once and for all. But contrary to
Xerox, the design process isn't that complex, and if designed
properly, doesn't require the constant diddling Xerox implies.
Admittedly the required modem is complex, but for a good RF
designer, not that much more complex than an Ethernet transceiver
and occupies the same place in the architecture. Production
economies and special chips will get the modem cost down, just
like for Ethernets.
2) For a given piece of Coax, broadband systems get 5-10 times the
bandwidth out of the cable. Broadband systems multiplex in two
domains: time in each channel, and frequency within the cable.
This allows data, voice, video, and whatnot all on the same cable.
I know it is possible to put voice and maybe even video on Ethernet,
but you can get a lot more with 5 logical networks within the same
cable! Most of the modem designs are frequency-agile, so there
are lots of optimizations possible for either high-bandwidth
applications, or private subnets, again, all in the same cable.
3) On the other hand, Ethernet is CSMA/CD, while most wideband
systems are either pure CSMA, or CSMA/hopefully-CD. You have to
look closely and think hard to understand the real behavior of
the wideband systems, and the suppliers haven't been to helpful
giving away fine details to make these distinctions.
4) The other problem with Ethernet is its DC coupling. While
it will work fine in an electrically-quiet office, there are
some very real problems with electrical noise, ground loops,
sheild currents, and other similar evils. Some are just nusiances,
but others can be deadly, if accidently applied to a human.
When wiring a large building, you won't run an Ethernet through
the building core with the power feeds and elevator circuits.
Here at LBL, we have power transients which would certainly kill
an Ethernet going further than one building, and possibly even
within a single building.
Ethernet and Broadband aren't natural enemies; they are in fact
more similar than different. Same basic algorithms, different
encoding on the medium. I forsee large building or campuses with
wideband backbones interconnecting small baseband subnets.
I think Ethernet will be viable in that many people are building
things to plug into it, but Broadband does have advantages for
large building or campuses. The two will co-exist quite usefully.
At least one vendor, Ungermann-Bass, intends to make sure of it.
-Mike
-----------------------------------------------------------------
gopher://quux.org/ conversion by John Goerzen <
[email protected]>
of
http://communication.ucsd.edu/A-News/
This Usenet Oldnews Archive
article may be copied and distributed freely, provided:
1. There is no money collected for the text(s) of the articles.
2. The following notice remains appended to each copy:
The Usenet Oldnews Archive: Compilation Copyright (C) 1981, 1996
Bruce Jones, Henry Spencer, David Wiseman.
