Hi Guy, all,
On your slice 4 – chain model
request, IMHO some comments are required. The diagram assumes simply e2e
reservation request forwarding, while local reservation computations are
performed in background, independently. It may be valid, but e.g. for AutoBAHN
it’s false. This calculation is done before forwarding the reservation request.
A results of this calculations (in the form of constraints, e.g. available
VLANs range) is attached to forwarded request. This allows domains to improve
their local reservation computation, taking into account predeceasing domains
information. Also in case a domain has no resources for particular reservation,
there is no need to forward the request. RSVP model (I would be glad if someone
could confirm that) works also in similar way, expecting Network (or node) to
validate resources, before forwarding the request to next Network element. I
think we should consider both scenarios, despite of the decided final functionality.
Best regards
Radek
--
____________________________________________
Radoslaw Krzywania
radek.krzywania@man.poznan.pl
+48 61 858 20 28
Network Research and Development Department
Poznan Supercomputing and Networking Center
____________________________________________
From:
nsi-wg-bounces@ogf.org [mailto:nsi-wg-bounces@ogf.org] On Behalf Of Guy
Roberts
Sent: Tuesday, June 23, 2009 2:44 PM
To: NSI WG
Subject: [Nsi-wg] Multi-domain message sequence
I have prepared an analysis of message
flows for multi-domain reservations –see attached presentation.
I think this raises some interesting
questions (see issues section) which I would like to discuss in tomorrow’s
call.
Guy
------------------------------------------------------------------------------------------------------------------
Guy
Roberts, Ph.D
Network Engineering
& Planning
DANTE - www.dante.net
Tel: +44 (0)1223 371 316
City House, 126-130 Hills
Road
Cambridge, CB2 1PQ, UK
------------------------------------------------------------------------------------------------------------------