> If I can continue to achieve those without manually configuring the PETR > RLOCs at the ITRs, we have what we need. But I didn’t see a way to do this in > the framework of the current spec.
You know I solved this problem in another way. By using a default mapping entry. Here is the idea (and what I implemented): (1) You have something register 0.0.0.0/0 and 0::0/0 with a large set of PETRs as RLOCs. (2) Those PETRs also run LISP-NAT according to the Interworking RFC. (3) ITRs load-split traffic to non-EIDs across the large cluster of PETRs. (4) The PETRs, after decapsulation, translate the source EID to its RLOC per Intewroking RFC. (5) Returned packets to the EID, come to the PETR which the ITR chose (which is also a PITR). (6) THERE IS NO BGP route injection, so there is no participation required by the underlay. (7) There is path symmetry (for good or bad). You don’t need negative prefix calculation here because anything more specific is either a LISP site or the default mapping. I have this running in a AWS VM, if anyone wants to try it out. Something to think about, Dino _______________________________________________ lisp mailing list lisp@ietf.org https://www.ietf.org/mailman/listinfo/lisp
