RE: reminder: things todo in august and shortly thereafter

[neil.2.harrison@bt.com]

> Venkata,
> 
> >> the example, after a transit LSR recognizes that 
> requirements are not met
> on
> >> the next hop (e.g., lack of sufficient available 
> bandwidth), the transit
> LSR
> >> sends crankback to the ingress LSR.  The ingress LSR then 
> selects another
> 
> > Why should the crankback reaches till ingress LSR. Any intermediate
> > LSR which has sufficient information to route the call/setup 
> > can handle crankback well enough (if the ero is not strict but
> > aggregated and handled by ABRs for example in MPLS hierarchical TE)
> 
> As specified in the I-D on crankback
> http://search.ietf.org/internet-drafts/draft-iwata-mpls-crankb
ack-01.txt,
the transit LSR can either crankback to the ingress LSR or a border LSR.  We
gave a simple example where it sends crankback to the ingress LSR; there are
other cases.

Jerry Ash
NH=> One (good) reason to crankback to the source is to spread the
restoration load away from (instead on concentrating it near) the point of
failure.
Neil