🚀 go-pugleaf

>From Odonnell@Yale Wed Mar 10 13:48:16 1982
Several points should be made:

First, about broadband.  Current RF modems have problems with a cumulative-noise
phenomenon; 40 dB down is not necessarily good enough if you plan hundreds of
modems.  Also, as I understand it a broadband cable is almost, but not quite,
many logically independent networks.  Problems arise on collisions: transmitters
beat against each other, spraying noise across all frequencies on the cable.

Second, cable bandwidth.  Ethernet provides one channel, with well analyzed
and understood behavior under wide-ranging loads.  The CATV systems partition
cable bandwidth in ways that may not correspond to the load; also, many of them
use questionable FDM and packet-allocation practices that look nice in theory,
but may match real loads quite poorly.  Comparing 'raw' bandwidths is
quite misleading.

Third, DC coupling.  Properly designed transcievers provide ground isolation.
Lighting has struck near PARC several times without problems (problems
DID occur once when cable ground was not properly isolated from building
ground; this is equally dangerous and/or likely with broadband systems).
No argument here between Ether/Broadband makes sense, unless the 'broadband'
medium is fiber optic.

Our view is that indeed, some mix of ether and broadband makes a great deal
of sense in planning for a large campus: in-building Ether distribution linked
via broadband gateways (the Ungermann-Bass systems look attractive for this).
-------