ietf
[Top] [All Lists]

Re: LISP is not a Loc-ID Separation protocol

2011-11-04 07:03:58
----- Original Message -----
From: "Templin, Fred L" <Fred(_dot_)L(_dot_)Templin(_at_)boeing(_dot_)com>
To: "Noel Chiappa" <jnc(_at_)mercury(_dot_)lcs(_dot_)mit(_dot_)edu>; 
<ietf(_at_)ietf(_dot_)org>
Sent: Thursday, November 03, 2011 3:54 PM
-----Original Message-----
From: Noel Chiappa [mailto:jnc(_at_)mercury(_dot_)lcs(_dot_)mit(_dot_)edu]
Sent: Thursday, November 03, 2011 7:43 AM
To: ietf(_at_)ietf(_dot_)org
Cc: jnc(_at_)mercury(_dot_)lcs(_dot_)mit(_dot_)edu

    > From: "Templin, Fred L" 
<Fred(_dot_)L(_dot_)Templin(_at_)boeing(_dot_)com>

    > one to one correspondence with the end system's multiple VPN
    > connections. The internal virtual interfaces keep the
VPNs separate.

As logically separate sources for incoming/outbound packets,
they are just
like multiple real interfaces. The name used for neither is really the
'identity' of the host, the names applied to those things
just identify
sources/sinks of packets.

    >> Suppose I assign two endpoint identifiers to a host.
Which is the
    >> host's identity?

    > Neither - that's the point. The host's identify is not
bound to any one
    > or multiple IP addresses.

I said "endpoint identifiers" (under whatever definition of
that term one
wishes to use), not "addresses".

OK, so then let's consider a related analogy. It is
not uncommon for a person (e.g., John Doe) to have
dual citizenship with countries A and B, i.e, John
"interfaces" with both countries. John would receive
a separate taxpayer identifier from each of A and B,
which has relevance only within the respective
country's "routing system". But, John's identity is
neither A nor B; John's identity is John.

Not necessarily so.  I agree with you so far, but the question is, when John
contributes to this list as AD and then as individual, is that one identity or
two?  Arguably, it is two, each of which could have multiple identifiers.

Likewise, a host can be differentiated from the applications running on it, and
in a large server farm, the relationship can be m:n, so there is an application
identity, with multiple identifiers, which is distinct from a host identity,
with multiple identifiers.

Identity is a slippery thing and much of the discussion on RRG, which I
faithfully followed, seemed to me to be a lack of agreement on identity, not
about identifiers.

That said, no way is LISP a separation of Locator and ID:-)  It seems to me a
bit like countries which included 'Democratic' in their identifier, which always
make me want to see some proof that they were indeed democratic:-)

Tom Petch


So, a single host can have multiple identities (whether one
does so via
multiple interfaces/interface addresses, or endpoint identifiers). So?

No; not multiple identities. One identity; multiple
interfaces and multiple addresses. Same as for the
taxpayer ID analogy.

Thanks - Fred
fred(_dot_)l(_dot_)templin(_at_)boeing(_dot_)com

Noel

_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www.ietf.org/mailman/listinfo/ietf



_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www.ietf.org/mailman/listinfo/ietf