[RAM] Re: LISP-01 I-D

[Robin Whittle <rw at firstpr.com.au>]

My understanding of LISP-01 is that LISP-00's IP-in-IP
encapsulation is not used at all, but that the original packet
becomes part of a new UDP packet.

The UDP packet's contents begin (page 12) with a "LISP header",
but this is not a header in the sense of an IP, TCP or UDP header.

UDP packets are also used for messages between ITRs and ETRs.
Similarly formatted messages are used with TCP, to the same port
number as the UDP messages, for the LISP-CONS system.

After reading LISP-01 I think that Ivip, as I am developing it,
has less and less in common with LISP-00.  Now that LISP-01 has
adopted UDP instead of IP-in-IP - and since I think IP-in-IP is
just fine for Ivip - there is even less in common.

What Ivip takes from LISP is some vital basic concepts:

  Using existing IP address space in two ways - some addresses
  as "Ivip-mapped addresses" ("EIDs" in LISP) where packets
  addressed to these are tunneled to some other IP address (an
  "RLOC" in LISP).

  A large number of ITRs to do the encapsulation.

  ETRs near wherever hosts with Ivip-mapped addresses are
  located.

  Some kind of push and/or pull global distributed database
  to control the operation of ITRs.

  The aims of providing portability of addresses, down to
  single IP address granularity (at least for IPv4) and
  multihoming.

A list of differences between LISP and Ivip would be much longer.

Probably in late August I will be able to create one or more Ivip
I-Ds with more fully-baked ideas - with a more cohesive
presentation.  I will continue to write up my ideas for this list
for those who are interested in reading fresh ideas.  This is
primarily to help people who are devising their own schemes and
want to copy or improve on my ideas - or to avoid them!  I see
this as the brainstorming phase, with this RAM list the place
where we throw ideas into the pot, combine them in different ways,
critique and improve upon them etc.

Some of what I write may be applicable to some LISP variants - but
I am concentrating on what I think are more expansive goals, with
a more incrementally deployable scheme, rather than trying to
contribute directly to LISP.

Hopefully writing these ideas up on this public web-archived
mailing list will mean anything which might have been patentable
will no longer be so.  If even one of my ideas winds up being part
of the final system, my efforts will have been worthwhile.


  - Robin     http://www.firstpr.com.au/ip/ivip/



_______________________________________________
RAM mailing list
RAM at iab.org
https://www1.ietf.org/mailman/listinfo/ram