Hi Gyan,

Thank you for your comments.


> This is an interesting idea and I support the concept of being able to 
> provide a TE cSPF constraint for KPI  like power consumption based green 
> routing and power save idea to power down ports or components not in use.  
> 
> Great work.  As this is cSPF path computation related I believe that LSR is 
> the appropriate WG.
> 
> Would this new power membership group constraint be applicable to segment 
> routing SR-MPLS & SRv6?


Could it be used by a controller supporting SR-MPLS or SRv6?  Absolutely. The 
information is there in the LSDB for anyone to use.  How should they use it? 
Well, that’s up to the folks writing the controller.


> I see RSVP-TE mentioned once but no mention of SR.


We haven’t gone there yet, but of course, it’s probably doable. There are 
signaling parts that are obviously not supported without RSVP.  Someone would 
need to provide that protocol work.  That part would probably fall under the 
domain of TEAS.


> If so please update the draft for SR use case of this feature.


Thank you for the suggestion.


> As mentioned by others operational considerations should be added for RSVP-TE 
> FRR, LFA, RLFA, TI-LFA make before break backup tunnels.  Also operational 
> considerations for any TE policy both RSVP-TE non FRR or SR policies where 
> backup paths could be impacted by power save modes. 7.2.2 and 7.2.3 should 
> account for all the FRRs above mentioned in putting an FRR backup path in 
> power sleep mode.  RSVP-TE FRR would have a double book so a book on primary 
> and backup path so would the sleeping bandwidth have accounted for the FRR 
> backup paths?   


Again, we are not standardizing how this is to be used. Ensuring the right 
level of redundancy for a given network and demand matrix is network specific. 
We can add some vague generalities, but they amount to “Do not aim weapon at 
foot.”


> Would there be some threshold configurable I am guessing that based on the 
> unidirectional sleeping bandwidth calculation, the resulting number if below 
> a threshold could then get the sleeping bit set.


There are many implementation alternatives.  I suggest you have a private 
conversation with your respective vendors.


> Also I wonder if an option or POP TLV for green DC energy footprint KPI 
> versus a DC running on commercial power using fossil fuels.  That would allow 
> routing for green DC POP based routing.  


As the GREEN WG is trying to prove, you can easily make this problem 
arbitrarily hard. We are trying to keep things simple for now. If it becomes 
compelling, things can always be extended.

T


_______________________________________________
Lsr mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to