Hi Noel,

I wrote too quickly that I thought of your critique:

>   too much focus on minor, passing problems, and not enough
>   attention to unavoidable architectural limitations

but actually, I think you wrote about problems which are significant.

I think they are inevitable consequence of making the wrong
architectural choices - which seems to have been due to strongly held
beliefs that it is impractical to have local full-database query
servers and real-time delivery of mapping to them.

Yet, as far as I know, the LISP folks at Cisco played a direct role
in designing many of the routers which make the Internet such a fast
and reliable global packet delivery system!  It takes less than 200ms
to get a packet from almost anywhere to almost anywhere else - and
the great majority of packets are delivered just fine.

If local query servers get real-time mapping, its not hard to have
them reliably and instantly alert any local ITRs which are caching
mapping for some EID prefix whose mapping has just changed.  Then the
end-users control the ITRs directly and they can do reachability
testing and ETR decisions etc. however they like, without relying on
anything built into ITRs or specified in LISP itself.

  - Robin
_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg

Reply via email to