km@emory.UUCP (Ken Mandelberg) (12/19/84)
Here is a description of our Vax configuration, and how we want to
connect them. It is rather a rather a long story, so don't readon
in less you are interested in building bastard ethernets.
1) Our central computer is just installing a short ethernet
connecting 3 Vaxen and an IBM machine in their computer room.
About 75 meters of cable.
2) Our department has its own Vax, which lives about 700 meters
from the computing center. We wish to both connect to the
central ethernet, and add a few local workstations. Our needs
would be met with a short ethernet connected to our Vax
(maybe 100 meters worth), plus a connection to theirs.
3) Right now we connect our Vax to one of the central Vaxen
by a point to point link. This uses a DMR11 and 700 meters
of RG50 75-ohm triax cable which was strung on telephone polls
at great expense. Everything talks TCP/IP so in principal the
current point to point link will give us use of the computing center
ethernet by routing through a Vax. However, this does not help
us add the local workstations. It is also puts extra load on an already
overloaded computing center Vax just for acting as a gateway.
4) Here are some monetary axioms. We don't have enough money to
install new ethernet cable to the computing center. We are
out of unibus slots on our Vax, so we cannot add an additonal
ethernet card without replacing the DMR. We could probably sell
the DMRs and afford a replacement ethernet card.
What our computing center suggested is that we do some impedence
matching and use the triax like it was ethernet cable, and join
the whole business into one 75+700+100=850 meter ethernet segment.
I know this violates at least the 500 meter per segment rule,
and I don't understand the electronics of the impedence matching.
I suppose we could put remote repeaters between the segments.
But there is still the problem of the 75 ohm cable. I also get
the impression that remote repeaters are expensive.
So finally, the question is this. Does the plan to build the 850
meter segment spliced together with nickle and dime components
have a chance of working? Does anyone have a cheap alternative?
--
Ken Mandelberg
Emory University
Dept of Math and CS
Atlanta, Ga 30322
{akgua,sb1,gatech,decvax}!emory!km USENET
km@emory CSNET
km.emory@csnet-relay ARPANET