[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PCN] Piggybacking -- a new edge behaviour

Please see below.

philip.eardley at bt.com wrote:

Hi

I think what Michael says basically agrees with what tom says (?)

For admission control there seem to be 2 cases, basically as tom says:

- where pcn signalling fate shares with existing resource signalling, eg
by "piggybacking" in the RESV msgs
- where pcn signalling doesn't fate share with existing resource
signalling, eg a special pcn msg carries pcn info. 
They key difference is that the first assumes there is an existing
resource signalling mechanism, whilst the second needs reliability on
the special pcn msg

Tom, I don't see you difference (2) as significant - I think that the
admission decision can be made at the ingress or egress for either [the
case of centralised decision point makes no difference; if you want to
fate share pcn signalling & resource signalling, then clearly the latter
must go through the centralised node]
[PTT] No. You are confounding the per-flow decision with the reporting of an admission state that conditions that decision. The actual decision whether to admit a specific flow can be made at the egress node only when resource signalling is being used. Otherwise it MUST be made elsewhere. 


for Flow termination
Signalling of pcn info can use the existing resource signalling - if the
latter allows asynchronous msgs. So for resource signalling protocols
that don't support this, then you have to use a special psn msg 
Best wishes,
phil

...