> One additional comment: > 4. I have seen assumptions that Map server and proxy to Non-LISP are on the > same box. It is a potential bottleneck in the future.
That is a deployment decision and not an architectural requirement. > DHCP would deliver only 1 Map Server IP to RLOC. Hence all overlays would > share the same Map Server. Not true. The map-server(s) are configured in ETRs. They can be distinct addresses or anycast addresses. > Proxy is a router with a low capacity CPU. It could become a bottleneck if > LISP implementation would become big enough. The PxTR needs to be built (by equipment vendors) like any other type of router and the capacity depends where it is deployed and the speed of links and aggregation of traffic comes to it. Clustering solutions make PxTRs scale. > It is better to assume initially the general case that Map Server and Proxy > are different boxes (not collocated). Yes, it is assumed by default and co-locating it is a deployment decision. Dino _______________________________________________ lisp mailing list [email protected] https://www.ietf.org/mailman/listinfo/lisp
