The Routing Over Large Clouds Mailing List Archive by date

Cell Relay Retreat>List Archive>month:1995-Mar> msg00035



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

Application Statement

  • From: "j.garrett" <jwg@garage.att.com>
  • Date: 9 Mar 95 02:04:00 -0500
  • Cc: rolc@maelstrom.timeplex.com
  • Original-From: garage!jwg (j.garrett)

Joel,

After some more thought I understand your point.
There is no guarantee that routing will determine
that the next-hop to a next-hop is the next-hop.
That is, routing could determine that router R1
is the next-hop to some destination, and also determine
that router R2 is the next-hop to router R1.
In this case an NHRP request would be sent to R2
and this NHRP transaction would resolve the address
of R1.  Therefore, my assertion in a couple of postings
to the ROLC list that NHRP can not do address resolution
of a next hop determined by routing was wrong.

I do hope that if routing is able to use router R1 as a
next-hop, that routing (and address resolution) will be
clever enough to send packets addressed to R1 directly to R1,
without going through some other next-hop router.

This said, I will offer one observation.  If you
want routing to determine which IP prefixes can be
reached without going through a next-hop router,
and don't want the routing decision to be made
independently for each host address, and if you
want an address resolution procedure that will
resolve a host address, the address resolution procedure
better not depend on sending a query to the (non-existant)
next-hop identified by routing.  Moreover, if routing has
learned from a routing peer that the host can be reached
without going through a next-hop router, the routing
peer that provided this information is a good
candidate to help resolve the host's address.

John Garrett
!h