The Routing Over Large Clouds Mailing List Archive by date

[Joel Halpern <jhalpern@mako.newbridge.com>]

I see that a new version of the NBMA ARP spec has been posted to the
repository.  As a participant, I have a few questions for other people in
this group:

1) We had discussed having a "trace" mechanism so that
    A) The response could follow the outbound path back, without the
	intermediate NAS having extra state,
    B) The requestor could place more trust in the response,
    C) THe alternate addresses could be used to cope with policy restrictions

2) We had discussed having a couple of bits to indicate whether the requestor
    or responder were routers.  The hope here was that by having the bits,
    the protocol might be extensible to use by routers in the future.  And in
    the mean time, routers would know how to ignore unsafe answers  (or
    even produce and use them, as a matter of local policy).
    The risk here is taht we are wasting bits because creating fields whose
    use is not well defined is the IP TOS trap.

3) Extensibility:  This has not been discussed as much in the group.  Given
   the uncertainty about the future of this problem space, it would be nice
   if the packet format was extensible, so taht we could carry more
   information in the future.

I am sympathetic to Juha's goal of getting something useful to solve the
simple problem quickly.  I would like other people's comments on these
specific capabilites.

Thank you,
Joel M. Halpern				jhalpern@newbridge.com
Newbridge Networks Inc.