The real problem

I would suggest that the number of prefixes in the global
routing table *is* linearly related to the amount of
computational overhead.

- paul