Instance-IDs are there (in the LCAF draft) to support LISP-VPNs. There is 
nothing in the main specifications because VPNs are out of scope for the LISP 
WG.

Dino

On Oct 23, 2011, at 1:53 PM, Joel M. Halpern wrote:

> The problem I have with that view is that although it is accurate, it is also 
> bascially not explained.
> For charter reasons, we left out any explanation of the private spaces, other 
> than putting in the hook for instance-ID.
> 
> So, yes, as long as all of the users of a given mapping system and given 
> instance ID use 10/ space in a consistent fashion for allocating IDs, it will 
> work.  But I don't see how a reader could understand that from the document.
> 
> Yours,
> Joel
> 
> On 10/23/2011 3:30 PM, Dino Farinacci wrote:
>>> By the way, it is interesting that Ari thought the EIDs would come out of 
>>> 10/8 space. That is obviously wrong, but maybe this misunderstanding gives 
>>> us a datapoint on the issue that I raised earlier about using the 10/8 
>>> addresses in examples.
>> 
>> Ah, but they could. That is the point. Since they are not injected in the 
>> core, an EID-prefix in 10/8 space can be made unique to both the mapping 
>> database and the data-plane core by coupling it with an instance-ID. And the 
>> instance-ID need only be unique *per mapping database system*.
>> 
>> Dino
>> 
>> 
>>> Jari
>>> 
>>> _______________________________________________
>>> lisp mailing list
>>> [email protected]
>>> https://www.ietf.org/mailman/listinfo/lisp
>> 
>> _______________________________________________
>> lisp mailing list
>> [email protected]
>> https://www.ietf.org/mailman/listinfo/lisp
>> 

_______________________________________________
lisp mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/lisp

Reply via email to