Hi Dino,

The question of separate namespaces for EIDs and RLOCs is absolutely
crucial to the difference between LISP and HIP.  HIP requires a
separate, new, namespace for EIDs and so cannot be introduced without
altering hosts and (I am not sure) perhaps applications too.

LISP could use separate namespaces, but for introduction on a
voluntary basis for solving the routing scaling problem, it does not
use separate namespaces (according to the only definition of the
term I can find, and no-one has offered an alternative on this list).

LISP EIDs are a subset of the global unicast range of addresses and
RLOCs are either another subset, used just for ITRs and ETRs, or
are all global unicast addresses which are not EIDs.

This means that LISP is potentially practical for introducing on a
global basis without upgrading all hosts and/or DFZ routers.

LISP *potentially* has separate EID and RLOC namespaces, but that
could only be useful in the distant future if there was either an
upgrade to all hosts for a new EID namespace or an upgrade to all DFZ
and some other routers so the IPv4, IPv6 or some other future global
network supported addresses which used a new RLOC namespace.

Since "Locator Identifier Separation Protocol" seems to cover the
entire field of "Locator Identity Separation" while HIP is also
another Locator Identity Split protocol, I and some other people think
it is vital that the Charter mention this.

While LISP could, in the future, involve separate namespaces, the
fact that it can work fine with RLOCs and EIDs being separate subsets
of the one global unicast address range (which is a single namespace)
is what makes it potentially practical for solving the routing
scaling problem.


Assuming IPv4, and with the understanding that an EID is for a host in
an end-user network using LISP-mapped addresses, your descriptions
make sense to me and seem to accord with your I-Ds:

> An EID is a 32-bit address used for socket-id identification in hosts.
> It is also an interface ID since the host can be multi-homed and
> attached to the subnets the IGP is routing within the site. An EID is
> not injected into BGP routing towards the core.
> 
> An RLOC is the PA IP address assigned to the CE/PE interface of the LISP
> router that resides at the site.

Can you comment on whether Noel and Sam are right to think that an
EID address could also be used as an RLOC address, which I think is
at odds with the definitions in draft-farinacci-lisp:

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

    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.


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

    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.


  http://www.ietf.org/mail-archive/web/lisp/current/msg00365.html   SH

    What do you mean that no EID is also an RLOC?  It's not clear to
    me that in interworking cases this is always true.


  - Robin