Re: [RRG] On the Transitionability of LISP

[Christian Vogt <christian.vogt@nomadiclab.com>]

David:

>> The requirement for support on both sides implies that an "upgraded"
>> edge network using mapped IDs will no longer be reachable from "legacy"
>> edge networks that do not yet support the mapping.  This is a
>> disincentive for edge networks to adopt the ID/locator split mechanism
>> during an early transition stage.
> 
> 	This would seem to be incorrect. There is no reason that
> 	early adopters need withdraw their "legacy" routes from
> 	the DFZ until it makes sense. 

I do agree that, to remain reachable, upgraded edge networks would have
to use their old locator space in addition to the new ID space.  What I
am concerned about are the following two problems that arise from this:

  1. It defeats all of the benefits of the ID/locator split.
  2. It makes reliable address resolution infeasible.

Regarding problem 1:  The four main benefits envisioned for an
ID/locator split are the following.

  (i)   Enabling edge networks to route packets via arbitrary providers.
  (ii)  Reducing the network reconfiguration cost related to rehoming.
  (iii) Reducing the size of the global routing table.
  (iv)  Reducing the update frequency of the global routing table.

Now, if an edge network is forced to maintain its old locator space in
addition to new ID space, then /none/ of (i) through (iv) will be
satisfied.  On the contrary, network administration overhead will be
increased, and the global routing table will become larger.

Regarding problem 2:  A coexistence of locators and IDs implies that the
result of address resolution depends on the location of the resolving
host:  Hosts in legacy edge networks MUST obtain legacy locators, while
hosts in upgraded edge networks SHOULD obtain IDs.

If there was only DNS for address resolution, then this could be
achieved by setting up DNS servers in upgraded edge networks so that
they provide IDs, while having DNS servers in legacy edge networks
continue to provide locators.  Filters at the borders of upgraded edge
networks would prevent DNS signaling from leaking between legacy and
upgraded edge networks.

However, there is a plethora of application layer protocols -- such as
SIP or peer-to-peer protocols -- that exchange addresses and thus
perform their own address resolution.  Addresses in packets from all of
these protocols would have to be translated at edge network borders.
This seems infeasible.  [Pekka Nikander calls this problem the "referral
problem" in draft-nikander-ram-generix-proxying-00.]

> 	                  You could even have example.com
> 	(legacy) and example-new.com (or whatever). Or both,
> 	depending on what makes sense.

Hmm, this is a transition mechanism that is based on changes in user
behavior...

- Christian


--
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