Re: PQC and QOSPF

[Yao-Min CHEN <[e-mail]>]

Ken,

Good point. 

If dependency were one-way then we would probably be okay 
but what's troublesome is the cross-dependency between 
RSVP and path computation.  To break the dependency it seems one
could use a measurement based appraoch as mentioned by Jon.
Alternatively, it is proposed in PQC to take QoS path computation
away from routing to merge with RSVP. 

Yao-Min  


Kenneth CARLBERG wrote:
> 
> > Although still not quite sure about the traffic measurement
> > part (have yet to investigate what such a technique can do),
> > I pretty much agree that we should "evolve routing, signaling,
> > and admission control (and charging) all separately."
> 
> one aspect to keep in mind in the separation model is that time and
> resources are needed to coordinate the separate activities.  if routing
> schema (A) is dependent on signaling schema (B), then (probably) care
> might have to be taken for potential race conditions.
> 
> -ken