Re: [lisp] LISP EID Block Size

[Rene Bartsch <[e-mail]>]

Am 2013-10-31 17:31, schrieb Sander Steffann:
>> I'll explain one experiment and will be crystal ear about it.
>> 
>> (1) We want to have a LISP site send to both LISP sites and not LISP 
>> sites.
>> (2) We want an ITR to do mapping database lookups when the destination 
>> address is an EID.
>> (3) We want the ITR to forward packet s by doing a traditional FIB 
>> lookup when the destination is not an EID.
> 
> I understand what the technical idea is. There are already EIDs out
> there (including my own network) that are not in any special prefix.
> Therefore "address is in special prefix" != "address is an EID".
> Unless you break LISP for already existing sites I don't see how
> having a special prefix is going to help in (correctly) determining
> whether an address is an EID.

Consider every consumer gets a random PI-prefix out of the global 
unicast address space. Now you have two problems with a public PITR:

1. What prefixes does the public PITR have to announce? Where to get the 
list from? Scan all Map servers?
2. Billions of random PI-prefixes need millions or even billions of 
BGP-routes to be announced by the public PITR via BGP.

If there's one dedicated EID address space a public PITR only has to 
announce ONE BGP-route to himself.

Best regards,

Renne