RE: Differentiated Routing, not only plain rambo-SPF

> I am really surprised that the need for having several=20
> routing table is a big (scalability) issue:
> (BTW:Logically it takes separate routing tables, one per=20
> DSCP. But you may as well maintain one routing table
>  and search the next hop entry also based on the DSCP)
> IMO stateless DiffRout-Forwarding is more scalable than MPLS or RSVP.
  Yes diffserv forwarding is less stateful when compared to MPLS=20
or RSVP. But that has nothing to do with scalability problems in =
maintaining=20
multiple routing tables.=20
There are scalability issues wrt single routing table itself. People are =
still=20
working on scalable solutions for mimizing the overhead, wrt adding =
mechanisms=20
like accounting and other valued added measurement functions, which =
requires=20
processing of pkts at line rate. So i am guessing, scalablity associated =
with=20
multiple routing tables, will be a long shot.

See again you have started with diffrout, then it was changed to =
diffser-aware-TE=20
and now mechanisms for basic diffserv arch.=20

> However,if the size of the routing table is an issue, then=20
> why wasn't it an issue so far:
> Each ingress node needs to know the egress node before=20
> setting up the respective next hop entry in the routing table.
> So why doesn't the IP header contain a field for the=20
> destination router address. It would allow to keep the routing tables
> small: one single next hop entry w.r.t. each destination=20
> router would be sufficient.
> Of course, it would be rediculous to expect any change in=20
> IPv4. But why wasn't this an issue for IPv6 ?
> When IPv6 was born, I guess there have already been routing=20
> tables of respectful size.
> Can anyone tell me?
_______________________________________________
Irtf-rr mailing list
[email protected]
http://puck.nether.net/mailman/listinfo/irtf-rr