Routing does not use an exponential NP-Complete algorithm (like the

travelling salesman). The travelling salesman problem tries to solve the

cheapest way in which a salesman can visit every one of a set of cities.

In fact, routing can be done in order (n) with a bounded metric and

bounded distance, since it really is only trying to find the cheapest way

to get to a single destination.

The routing mentioned was optimal routing of virtual circuits.

It thought of as NP-complete.

There's quite a lot of science (commodity flows, and such) essentially

investigating efficient ways to find quasi-optimal solutions for the

problem.

What I don't know, is why is it that SS7, the telephone routing protocol,

can do some of the things that are required, like load sharing across

unequal paths, for example. Does anyone have any insight into this?

They simply precompute routing for different scenarios and then just load

it. Precomputation can take days at pretty big machines. SS7 is

no magic -- just a protocol.

--vadim