> From: Robert Raszuk <[email protected]>
> even those proposed just push the issue to a mapping plane
I did say that doing TE in a mapping would not be as good as an upgraded,
modern routing architecture (one suited to a global-scale information network
- but let's not get me started on that rant) that had TE capabilities built
into it... _but_ using the mapping plane is still much better than using
more-specifics.
Just like doing multi-homing by injecting routes into the global
destination-vector routing is just about the worst way to support MH, same
thing for TE. Every router in the entire default-free core has to dedicate a
routing table slot to each more-specific - and not just that, it's not just
table entries, there's also routing system overhead to maintain it (CPU, etc,
etc).
If you do it in a mapping plane the only people who have to pay the overhead
are the people _who actually talk to that destination_ - not everyone in the
whole world. Does this not sound like a fairer way to support this?
Noel
_______________________________________________
GROW mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/grow
