> From: Curtis Villamizar (curtis@workhorse.fictitious.org)
> Date: Mon Mar 26 2001 - 16:48:38 EST
> The big caveat in the paper is (in the authors terminology but
> paraphrased), the approach doesn't work for loopy topologies for some
> values of "work".
Sorry to jump on this so late but I only notice about this
disucssion through another discussion on the MPLS list.
The example topology you provided is not "loopy". One of
the issues in applying the results of the paper in practical
networks is that it may require unequal load-sharing
among equal-equal paths. If you accept that, the topology
you suggested can be handled.
Cheers
Zheng
This archive was generated by hypermail 2b29 : Mon Aug 04 2003 - 04:10:03 EDT