[lisp] Consensus? EID and RLOC use of the same address = separate namespace debate

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

Short version:   I see no evidence in the list messages that
                 rough consensus has been reached that "there
                 will be cases where the same IP stands both
                 as an EID and a RLOC".

                 Private discussions were mentioned, but AFAIK
                 these should have nothing to do with judgements
                 about rough consensus.

                 Since 2007-01 draft-farinacci-lisp-00 to 12 has
                 this as an absolute requirement:

                   EIDs MUST NOT be used as LISP RLOCs.

                 I am perplexed that I have so far been unable to
                 prompt anyone to write to this list agreeing
                 with this requirement.

                 I go further and state that within the context of
                 LISP being a practical solution to the routing
                 scaling problem, a particular IP address CANNOT
                 successfully be used as both an EID and RLOC.


Hi Sam,

In "Re: [lisp] My proposed revisions to the charter - LISP lacks
proper terminology":

   http://www.ietf.org/mail-archive/web/lisp/current/msg00336.html

you wrote, in part:

>> Sam wrote that a single address would "typically" not be used in
>> EID and RLOC roles.  For a practical LISP system, "typical" is not
>> strong enough - it is impossible.
>>
>> Noel still seems to think it is possible, but has yet to explain
>> how.
>
> I think there has been enough discussion on-list and other private
> comments that the rough consensus of the participants so far is
> that there will be cases where the same IP stands both as an EID
> and a RLOC.

I am completely unable to see how you as co-chair could decide that
rough consensus has been reached on this.

The messages so far are linked to at:

  http://www.firstpr.com.au/ip/ivip/namespace/

All of the following discussion is within the context of LISP being a
practical solution to the routing scaling problem, which means it
needs to work properly with unmodified hosts and unmodified DFZ
routers.

If LISP could work with the same particular IP address being used for
both an EID and an RLOC, then this could only occur due to the
devices which handle the packets differently according to which role
the address is being used in (ITRs and ETRs) have some way of telling
whether the address is to be interpreted as an EID or as an RLOC.
This would only be possible if there were in fact two separate
namespaces for RLOC and EID and if there was some mechanism by which
the ITRs and ETRs could recognise which namespace to use when
interpreting the address.

Without that, the an ITR will eat its own emitted encapsulated packet
or the encapsulated packet will get out to the DFZ and be forwarded
to a PTR, which will regard it as a traffic packet with an EID
destination address - to be encapsulated and tunneled to some RLOC
address.  I wrote about this specifically at the end of my last message:

  http://www.ietf.org/mail-archive/web/lisp/current/msg00335.html

Sam, I think that in stating that rough consensus has been reached
that "there will be cases where the same IP stands both as an EID and
a RLOC." you are also implying that rough consensus has been reached
that LISP (in the context of being a practical solution) does or can
involve two separate namespaces for RLOC and EIDs.

I see no evidence for such consensus.

I wrote extensively that:

  There can be no such separate namespaces.

  Trying to use a particular address as both an RLOC and EID
  cannot work.


Noel Chiappa asserted that there could be two separate namespaces:

  http://www.ietf.org/mail-archive/web/lisp/current/msg00274.html


Dino Farinacci wrote:

  http://www.ietf.org/mail-archive/web/lisp/current/msg00291.html

that the AFI field could be used to provide any number of namespaces,
but the example he gave could not be used by LISP if it was to be a
practical solution to the routing scaling problem.


Scott Brim wrote:

  http://www.ietf.org/mail-archive/web/lisp/current/msg00296.html

that it was impossible to prevent someone using a field for a
different role to what was intended.


Noel wrote:

  http://www.ietf.org/mail-archive/web/lisp/current/msg00332.html

    So if it's _possible_ to use the same bit pattern as both an EID
    and an RLOC, my guess is people will do it, no matter what the
    documents say. And my take is that, because of _other_ concerns
*   (e.g. limiting routing overhead), it will be technically
*   possible, which means people probably will do it no
    matter what we say in any documents.


That's it as far as I can tell.

Despite my repeated urgings no-one has described an example of how
two separate namespaces could be used in a practical LISP solution to
the routing scaling problem - or what amounts to the same thing, how
the one IP address could be used both as an EID and as an RLOC.


Sam, you may well have been part of private discussions I am not
aware of.  But as far as I can see, those discussions should have no
bearing on your decision about whether consensus as been achieved.


So far, no-one has produced anything more than assertions or
inappropriate examples when disagreeing with my argument.

No-one has pointed out where in the LISP I-Ds the concept of two
separate namespaces is specified.

No-one has pointed out where in the I-Ds this concept of separate
namespaces is allowed or where it is allowed for an address to be
used simultaneously as an EID or RLOC.

It is easy to quote chapter and verse on why an address can't be used
both as an EID and RLOC.

Since January 2007 this has been a part of the basic LISP I-D [!!!],
in a form which would constitute an "absolute requirement" in an RFC
2119 RFC.

  http://tools.ietf.org/html/draft-farinacci-lisp-12

      Routing Locator (RLOC):

         the IPv4 or IPv6 address of an egress tunnel router (ETR).
         It is the output of a EID-to-RLOC mapping lookup.  An EID
         maps to one or more RLOCs.  Typically, RLOCs are numbered
         from topologically-aggregatable blocks that are assigned
         to a site at each point to which it attaches to the global
         Internet; where the topology is defined by the connectivity
         of provider networks, RLOCs can be thought of as PA
         addresses.  Multiple RLOCs can be assigned to the same ETR
         device or to multiple ETR devices at a site.

   Endpoint ID (EID):

         a 32-bit (for IPv4) or 128-bit (for IPv6) value used in the
         source and destination address fields of the first (most
         inner) LISP header of a packet.  The host obtains a
         destination EID the same way it obtains an destination
         address today, for example through a DNS lookup or SIP
         exchange.  The source EID is obtained via existing
         mechanisms used to set a host's "local" IP address.  An EID
         is allocated to a host from an EID-prefix block associated
         with the site where the host is located.  An EID can be used
!!!!     by a host to refer to other hosts.  EIDs MUST NOT be used as
!!!!     LISP RLOCs.  Note that EID blocks may be assigned in a
         hierarchical manner, independent of the network topology, to
         facilitate scaling of the mapping database.  In addition, an
         EID block assigned to a site may have site-local structure
         (subnetting) for routing within the site; this structure is
         not visible to the global routing system.


I am really perplexed that I have so far been unable to get anyone to
 write to this list acknowledging that a given address cannot be used
both as an EID and an RLOC in any implementation of LISP which would
be practical for widespread voluntary adoption.

  - Robin