| Is the subject of this RG "routing for infinite bandwidth links", or
| "routing"?
The latter, of course. But if we consider QoS in the routing system,
then we have to be prepared for "infinite bandwidth links" whose
vertices DO NOT want to maintain or advertise lots of state about their
QoS capabilities.
I see no fundamental reason why QoS *must* be subsumed by the
routing layer, but do see some arguments about why it might be.
Since this can be done in *many* ways -- for example, different sets
of maps for different qualities of service -- there's no point
in requiring our routing system(s) to be QoS-ignorant, and
likewise, I don't think it's wise to require them to be QoS-savvy either.
Perhaps a pragmatic approach would be to assume some nodes are willing
to maintain and advertise state about different QoSes, while some
nodes are not.
Sean.
This archive was generated by hypermail 2b29 : Mon Aug 04 2003 - 04:10:03 EDT