The MPLS WG Archive

[Anouk Rocher <anouk@ureach.com>]

David,

Once again thanks for your very helpful email

[cut]
>These constraints all serve to generate an explicit route.  This ERO may
>be computed by the router, by manual configuration, or by some
>standalone management server.  Once the ERO is computed, RSVP doesn't
>have to care about the constraints that were used to compute it.
>
>-- David

I didn't state it clearly enough in either of my messages - yes I do
understand that typically paths based on costraints are calculated
at the LERs (not considering the management station case here) and
the path along with the path requirements fed into a signalling protocol
-  either RSVP/TE or CRLDP.

I was mostly trying to figure out how the calculate requirements are
mapped into the various RSVP objects . Specifically if the ingress LER
calculates (say) a 10Mbps path through a network and specified all the
hops to pass through via ERO, what else should it do to tell the nodes
along the path that 10Mpbs should be given to the LSP being set up.

>From your explanation, I understand that first the ingress LER sends
out a TSPEC saying that 10Mps is requested along with an ADSPEC
specifiying the type of service (say controlled load). The Egress
LER returns a FLOWSPEC. Each node on the return path then begins to
use the parameters in the FLOWSPEC to provide the service originally 
requested by the ingress and confirmed by the egress. 

First, Derived from a re-glance at the documents and your email again,
is this flow of events in the ballpark even if not exact?

Second, is it legal to for an intermediate LSR to use the TSPEC and ADSPEC
parameters on the forward path rather than the reverse path? I was thinking
of this in order to not lose available resources during the forward path
which may not be around when the messages start coming back from the
egress.

Anouk





________________________________________________
Get your own "800" number
Voicemail, fax, email, and a lot more
http://www.ureach.com/reg/tag