Path: utzoo!utgpu!news-server.csri.toronto.edu!clyde.concordia.ca!uunet!tut.cis.ohio-state.edu!zaphod.mps.ohio-state.edu!uakari.primate.wisc.edu!pikes!boulder!daemon
From: feldman@spiff.Tymnet.COM (Steve Feldman)
Newsgroups: comp.dcom.sys.cisco
Subject: Re: Routing thru X.25 from a cisco to a Sun?
Message-ID: <25073@boulder.Colorado.EDU>
Date: 23 Aug 90 09:12:44 GMT
Sender: daemon@boulder.Colorado.EDU
Lines: 35

We ran into a problem when we tried this a year or two ago.  There is a
philosophy difference between cisco and Sun on how IP routing over X.25
should be done.  cisco routers like to treat an X.25 network as a
single multi-point network, where each router has a list of X.121
addresses corresponding to the IP addresses for each other router, and
will build circuits on demand and close them after an inactivity
timeout expires.  (This is similar to the way DDN X.25 operates, I
believe.)

Sun's X.25 over IP implementation, however, likes to treat the X.25
network as a collection of permanent point-to-point links, each with
its own network or subnet number.  It also insists that one end of
each link be responsible for ensuring that the circuit is up at all
times.

We did get the Sun to talk to the cisco router, but we had these problems:

1) Since the cisco assumed that each router on the X.25 network
was directly connected to each other one, it refused to route an
IP packet back over the same X.25 interface it came in on.

2) The cisco liked to build parallel X.25 circuits to the same destination.
Sun's x25manager barfed at this.  I think there might have been a way to
disable this on the cisco, but I'm not sure.

3) The Sun had to take the responsiblity for keeping the circuit up,
since it doesn't do circuit building on demand.

We did these experiments over a year ago, so perhaps cisco has changed
their implementation to get around these problems.  I'm pretty sure
that Sun hasn't done anything about it.

	Steve Feldman
	BT Tymnet, Inc.
	feldman@Tymnet.COM