I am going to have to call bullshit on the MPLS fast reroute thing there Wayne. The canonical counterexample is Sprint.

To be sure, Sprint and Ebone have shown that it is possible to run a very high quality (low jitter) pure IP network, but...

Excellent engineering and ops folks top the list, followed closely by sufficient capacity,

This poses a bit of a problem. If we accept that these two things are absolutely necessary in order to run a pure IP network, i.e. no MPLS and no ATM or frame at layer 2, then that leaves out most of the world. By definition, excellent engineering and ops folks are the top minority in their field. If we assume that 10% of engineering and ops folks are "excellent", that means that 90% of network operators are unable to play this game. Also, not everyone has a facilities-based network where they can afford "sufficient" capacity. And last but not least, this whole discussion was about diversity and resilience. I would not be happy to see a network ecosystem in which every network was designed and run like Sprint's network. There is a place for MPLS and in the end, it may even prevail because of its ability to handle multiple best paths and to do fast reroute. The jury is still out on that. --Michael Dillon