Re: OK, shim6, if we must [Re: [RAM] Ivip (was ViP ...)]

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

Brian and Robin:

>> I think this
>> has to happen soon for IPv4, because the BGP system can't be
>> expected to keep up with the growth in the number of advertised
>> prefixes.  We have to find a way to provide portability,
>> multihoming, TE and ideally great mobility for millions of
>> end-users, without altering hosts.
>>
>> SHIM6 would still work with any such LISP or Ivip system for
>> IPv6 - so it is "orthogonal" to any such system.  However the
>> things which SHIM6 would provide would already be largely or
>> entirely available via the LISP/Ivip system.
> 
> So what? The approaches are orthogonal, as you say.

Shim6 and LISP/Ivip can be combined to allow a multi-interfaced Shim6
host to spread a connection across different access networks, while each
access network may use LISP/Ivip internally for multi-homing with
multiple providers.

Other than this, I don't see an orthogonality between Shim6 and LISP/Ivip.

>> and
>> does not require any new software or for the host(s) whose
>> addresses are being managed to be in any way aware of it.
>> Nonetheless, a single host could manage its own LISP/Ivip
>> mapping, or some centralised system could perform this task for
>> it, for instance to choose the best ETR in a multihoming setting
>> when one link fails.
> 
> Yes, that might be a way for users to take control back from ISPs,
> although I think shim6 will prove an easier way to do that.

Brian, I'm not sure what you mean by "taking control back from ISPs".
But in any case, once a multi-homed access network to which a host
attaches performs address mapping (through LISP or Ivip), the host can
no longer choose between different providers (or ETRs) of that access
network.  All the host sees is provider-independent addresses.

- Christian


_______________________________________________
RAM mailing list
RAM at iab.org
https://www1.ietf.org/mailman/listinfo/ram