Re: [RRG] Consequences of no renumbering...

[jnc@mercury.lcs.mit.edu (Noel Chiappa)]

    > From: "Tony Li" <tony.li@tony.li>

    > If we look at the current LISP transition plan, there is currently a
    > requirement for sites to renumber once to get into an aggregateable EID
    > space.

As others have said, this isn't an architectural requirement of any jack-up
scheme (such as LISP); indeed, when I first starting thinking about the
deployment of jack-up schemes, many moons ago, I was assuming no renumbering.
(I recall drawing some pictures and talking at length about what routing
tables would look like in various places during the deployment, implicitly
based on that assumption, but I'm not sure everyone followed what I was
ranting about! :-)

Yes, there are routing issues if you don't renumber, but depending on how you
deploy it (e.g. if you restrict yourself to having a contiguous chunk of the
network doing routing on 'locators' only) those issues can be reduced to
mostly second-order cost/benefit kinds of things (i.e. not 10:1 cost
variances). (Caveat: I haven't looked at the current LISP system for
distributing mapping delegations [ALT?], I'm only familiar with CONS, and
there may be some aggregation issues with that portion of the system too.)


    > If renumbering is not required, then EID space doesn't aggregate. If
    > transition boxes (PTRs) advertised EID space into legacy routing, then
    > it would imply that there wouldn't be any reduction in prefix count
    > until transition was wholly complete.

In looking at those pictures/messages I mentioned above (it was on the
architecture-discuss list, on Nov 10/11 2006, in the thread "Ok, so what
should we do then?" if anyone cares :-), it becomes clear to me that this
whole issue of 'how is the routing going to operate' is one that really needs
systematic exploration in depth (something I bitched about back then, too -
see the message 'It's the routing, people!' :-).

For instance, although I looked at both having only a single routing plane, in
which 'locators' and 'addresses' mixed (which seems to be the assumption in
some of what you're saying), I spent more time thinking about systems with two
separate routing planes (one for each), and where _not all routers
participated in both planes_ (exactly because, as you point out, otherwise
"there wouldn't be any reduction in prefix count [in any router] until
transition was wholly complete"), examining such issues as 'where, and how,
would routes from one need to be injected into the other'.

I don't propose to do such a systematic exploration in depth in this message,
but I will once again plead for such a systematic exploration - at an
abstract level, to make the fundamental issues clear - of how the routing
would work during the deployment phase (which is certain to be lengthy).

	Noel

--
to unsubscribe send a message to rrg-request@psg.com with the
word 'unsubscribe' in a single line as the message text body.
archive: <http://psg.com/lists/rrg/> & ftp://psg.com/pub/lists/rrg