Hi Tony, Your reply makes it very clear what should be the case and confirms what I was hoping for. I only asked based on this sentence of yours:
"If an interface is not power sleep capable, then its PSP is not relevant in metric computation." I did not know what "metric computation" you were referring to. Was this normal SPF - hope not ? Was this CSPF on headends - hope not ? Was it something else ? I asked to make sure metrics of all links are relevant in both SPF and CSPF computations. Many thx, Robert On Sat, Aug 29, 2026 at 7:30 PM Tony Li <[email protected]> wrote: > > Hi Robert, > > > > But the way I read what you said was different. I assumed that 100% of > links must advertise power related information in order for them to be > eligible and become a player in an TE computation game. > > > > I hope this is not the case. > > > We are clearly miscommunicating. Let me try again. > > All links are considered as part of TE, including sleeping links. To > ensure that there is always a redundant network, only some links are power > sleep capable. For the ones that are power sleep capable, we consider the > PSP in computing the path metric. This causes traffic to condense on links > that are not power sleep capable. > > I hope this is clear. > > An operator COULD mark all links as power sleep capable, but this is > unlikely to produce a satisfactory result. You could, for example, produce > a topology that is a tree, and non-redundant. Thus, we advise operators to > not do this. Rather, determine a minimun redundant topology, leave those > links as not power sleep capable, and that will act as a lower bound for > the network. > > Cheers, > Tony > > p.s. This is very much into the TEAS draft, not the LSR draft, yet we’re > not on the TEAS list… > >
_______________________________________________ Lsr mailing list -- [email protected] To unsubscribe send an email to [email protected]
