Hi Tony, Hmm so you are saying that you are proposing power optimisation architecture without central controller which will have a global view of all demands in the network and will reduce the paths of the packets ?
I am not seeing how can you do that safely with fully distributed and independent CSPF running on each ingress only knowing his view of the network. Moreover as traffic is often bi-directional doing this independently in both directions makes it even more fun to watch. And sure I am personally not a big fan of controllers too, but when it comes to global view of the network I have not seen an alternative for any type of optimization. Kind regards, R. On Sat, Aug 2, 2025 at 5:04 PM Tony Li <[email protected]> wrote: > > Hi Robert, > > Thank you for your comments. > > > > 1. Do we really see that carrying power consumption for the power groups > is something that belongs in link state protocol and flooded to all nodes > vs something that should be done with streaming telemetry to power > optimizing controller ? > > > Yes, given where we are. Many networks today continue to avoid the > controller approach. Several have tried that path and found it > unsatisfactory. Pragmatically, those who make it work seem to be networks > that have a large programming staff and are willing to devote many > resources to the care and feeding of the controller. They are very much in > the minority. > > We can have this debate, but it’s pretty off-topic for LSR and will not > change reality. > > > > 2. Why does the subject draft consider only RSVP-TE when reporting > "Sleeping Bandwidth" ? Besides wouldn't allocated by RSVP-TE link bandwidth > simply time out after 15 sec when refresh is not seen on the "sleeping > link" ? > > > We would be happy to consider other signalling mechanisms if you like. The > point is to move traffic off of links and then let them sleep and later > restore links when capacity is required. The goal, of course, is to do this > without noticeable packet loss. > > > > 3. To accomplish your overall objective I see authors lean towards path > computation as the only way. How about a completely different approach > based on demand vs capacity (possibly with queuing) monitoring and > graceful link, line card draining and natural link/line card bypass via SPF > without any necessity of CSPF ? > > > We await the Internet draft with the details of your proposed > architecture. Until then, power conservation seems to suggest that > effective traffic consolidation is an excellent tool for network power > reduction. > > Tony > >
_______________________________________________ Lsr mailing list -- [email protected] To unsubscribe send an email to [email protected]
