On Tue, Dec 18, 2001 at 05:49:07PM -0500, Kastenholz, Frank wrote:
>
> The current _tools_ for implementing "policy" are what
> Vince describes. But we're not trying to require the current
> tool set (if we do, we probably would end up with the current
> system, but maybe with 48-bit AS numbers :-). We are trying
> to figure out what it is that the operators are actually
> trying to do and then say "the routing system must support
> that" (at least, insofar as it is relevant).
>
There are dangers inherent in such an approach. First, it places
somewhat arbitrary restrictions on the allowed solution which may
prevent significant improvements. Second, I suspect there are
things "operators are actually trying to do" that are defined by
the limitations of the current tools, not what would be desired
without those limitations.
> But this is a very hard problem.
>
I think that is agonizingly clear.
Ben
This archive was generated by hypermail 2b29 : Mon Aug 04 2003 - 04:10:03 EDT