Hi Ran,

Can you explain what you mean by Q2, and what you think are the
logical implications of agreeing with both statements?

Regarding your questions:

> A) "The Internet continuing down the current architectural path,
>     whereby site multi-homing increases the size/entropy of the 
>     DFZ RIB/FIB is not believed to be scalable or viable."
>     [  ]  YES
>     [  ]  NO

I understand this clearly and agree - as I guess everyone else does.

> B) "There is no reason to believe that a scalable solution for
>     site multi-homing will appear in the future so long as the 
>     Internet proceeds with current architectural approach to 
>     site multi-homing."
>     [  ]  YES
>     [  ]  NO

This may be nonsensical.  The condition is:

  so long as the Internet proceeds with current architectural
  approach to site multi-homing

where I guess "proceeds" means continues to use the current
unscalable approach (multihoming end-user networks advertising their
individual PI prefixes in the DFZ), either exclusively or at least as
much as now, with the potential addition of some other approach.

The the question is whether, under these circumstances, in the future:

  a scalable solution for site multi-homing will appear

If "appear" means "be adopted widely enough to make a difference"
then that can't happen with the condition of "proceeding", if
"proceeding" means "using exclusively".

If "appear" means that an architecture will be designed, at least to
the stage where people could imagine how might work, then I would say
this has already happened:

  CES architectures:  LISP and Ivip could solve the scalable routing
  problem, for IPv4 and IPv6 - though I reckon Ivip could do it
  better.  Both could also support the TTR Mobility system for IPv4
  and IPv6.  Maybe IRON-RANGER could solve the routing scalability
  problem too, though I think some of its mechanisms are IPv6-based.

  The CEE (AKA Locator / Identifier Separation) architectures:

     ILNP - Identifier-Locator Network Protocol
     Name-Based Sockets

  could all solve the problem for IPv6 if they worked as planned.  I
  don't yet fully understand how they would work, but I suspect they
  would not be as good for Mobility as the TTR Mobility approach -
  which none of them support.  If any of these did work as planned,
  they could in principle solve the routing scalability problem in
  terms of portability, multihoming and inbound TE for non-mobile
  IPv6 networks.  However, this requires full adoption by all hosts
  in all networks, and the extra delays and burdens on hosts which I
  object to: (msg06250).

I don't think the continued usage of the current unscalable approach
has inhibited the development of new scalable routing architectures.

Nor do I think its continued use would inhibit the adoption of a
suitably attractive scalable routing architecture.  I think the most
attractive is Ivip, with LISP second.  I don't think there are any
other scalable routing architectures which would be attractive enough
to have widely adopted on a voluntary basis.

I think it is good seek agreement/consensus, but I will sit out this
vote because I regard the second question to be ambiguous, confusing
and/or nonsensical.

  - Robin

rrg mailing list

Reply via email to