Re: Single subnet

Vernon Schryver <vjs%rhyolite.wpd@sgi.com> Sun, 20 May 1990 00:42 UTC

Received: from merit.edu by NRI.NRI.Reston.VA.US id aa09216; 19 May 90 20:42 EDT
Received: Sat, 19 May 90 19:42:00 EST from ucbvax.Berkeley.EDU by merit.edu (5.59/1.6)
Received: from SGI.COM by ucbvax.Berkeley.EDU (5.63/1.41) id AA23055; Sat, 19 May 90 17:41:30 -0700
Received: from whizzer.wpd.sgi.com by sgi.sgi.com (5.52/900423.SGI) for @ucbvax.berkeley.edu:fddi@merit.edu id AA19811; Sat, 19 May 90 17:41:47 PDT
Received: from rhyolite.wpd.sgi.com by whizzer.wpd.sgi.com (5.52/900423.SGI) for sgi.sgi.com!merit.edu!fddi id AA14155; Sat, 19 May 90 17:41:44 PDT
Received: by rhyolite.wpd.sgi.com (5.52/900423.SGI) for @whizzer.wpd.sgi.com:deering@pescadero.stanford.edu id AA05431; Sat, 19 May 90 17:41:42 PDT
Date: Sat, 19 May 90 17:41:42 PDT
From: Vernon Schryver <vjs%rhyolite.wpd@sgi.com>
Message-Id: <9005200041.AA05431@rhyolite.wpd.sgi.com>
To: Steve Deering <deering@pescadero.stanford.edu>
Subject: Re: Single subnet
Cc: fddi@merit.edu
Status: O

> From: Steve Deering <deering@pescadero.stanford.edu>
> I presume that the FDDI MAC or PHY layer detects unwrap events.  Can they
> or should they not be signalled upward as "change of topology" events?

There is nothing in SMT, MAC, PHY, or PMD that reliably and quickly informs
anything about a distant (ie. at some other station) wrap<->thru transition.

Imagine you have a string of three MAC-less concentrators, with the middle
one down, and the ring wrapped at the other two.  When the middle one
enters the ring and the other two CMT-signal the WRAP away, who is supposed
to send any frames announcing the happy event to those machines with
their OSI stacks?

More prosaically, what if a pair of stations unwrap the ring between them,
but their SMT announcement frames get gobbled by the monsters that eat
tokens?  Eventually one might hope that all stations will figure out the
change, after a few multiples of the 30-second NIF rate, if you assume
that all stations are monitoring all NIF's, and not just the ones they are
required to.

Local topology changes are no big deal, since at worst you might have to
violate layering.  It is distant topology changes that trash ideas like
bridges between the primary and secondary rings, or single IP-network for
both rings.

I've been told by a bridge builder of hardware to detect wrap<->thru
transistions.  You can measure ring latency before and after every CLAIM,
and infer something by a sudden large change.  You can also look for your
own claim or beacon frames on the "wrong" ring.  This requires that you
tie both of your special MAC's together in a novel (ie. non-standard) fashion.

I'm waiting for someone to suggest that all dual-MAC stations be required
to have the extra hardware so I can send them a special greeting.


vjs