Hi, Mike:
Thanks for your suggestions. I have added the following sentences in section 6.2 as: "To accomplish ECMP effects, the PCE can send multiple EPR objects to the same node, with the same route priority and peer address value but different next hop addresses." Will update the draft later together with the comments from Susan. Thanks for your comments, would like to see your more suggestions on the solution. Best Regards Aijun Wang China Telecom From: [email protected] <[email protected]> Sent: Tuesday, July 27, 2021 8:41 PM To: Aijun Wang <[email protected]>; 'Mike Koldychev (mkoldych)' <[email protected]>; [email protected] Cc: [email protected] Subject: RE: [Pce] Follow up about my question on the mic Hi Aijun, Thanks for clarifying Q1. Regarding Q2, you can probably mention that in the draft. Multiple EPR objects MAY be sent for the same destination, which results in ECMP to that destination. Thanks, Mike. From: Pce <[email protected] <mailto:[email protected]> > On Behalf Of Aijun Wang Sent: Monday, July 26, 2021 10:55 PM To: 'Mike Koldychev (mkoldych)' <[email protected] <mailto:[email protected]> >; [email protected] <mailto:[email protected]> Cc: [email protected] <mailto:[email protected]> Subject: Re: [Pce] Follow up about my question on the mic Hi, Mike: Thanks for your questions. Please see the replies inline. If you have more questions based on the followings answers, we can discuss them accordingly. Best Regards Aijun Wang China Telecom From: [email protected] <mailto:[email protected]> <[email protected] <mailto:[email protected]> > On Behalf Of Mike Koldychev (mkoldych) Sent: Tuesday, July 27, 2021 6:36 AM To: [email protected] <mailto:[email protected]> Cc: [email protected] <mailto:[email protected]> Subject: [Pce] Follow up about my question on the mic Hi Authors, Just following up about my 2 questions during the PCE WG session about https://datatracker.ietf.org/doc/html/draft-ietf-pce-pcep-extension-native-i p-14. Question 1: Is every prefix going to be advertised (via the RR) to every node in the native-ip domain, even if those nodes are never on-path for that prefix? [WAJ] Yes, the behavior of RR is unchanged. If my understanding is correct, the "BGP peer nodes" (R1 & R7 in Figure 4) would receive the PPA (Prefixes) and would inject these prefixes into the RR. The RR would then flood these prefixes (as regular BGP IP routes) to every single node in the domain (R2, R4, R5, R6) with the next-hop being set to R1 or R7. Please let me know if my understanding is correct or am I missing something. So even though R5 and R6 in this example are off-path, they would receive the prefix route from the RR? [WAJ] Yes, R5 and R6 will also receive such prefix advertisements via the normal RR behavior. But on router R5&R6, the route to the BGP nexthop (R1&R7) is learned from the IGP protocol, not from the EPR(Explicit Peer Route) Object. Then after the recursive process, the forwarding path on R5&R6 will be along the normal non-optimal path. Question 2: Have you thought about ECMP/UCMP (Equal/Unequal Cost Multipath)? How would you implement it in your protocol? [WAJ] Currently, we consider only the ECMP application. If PCE wants to deploy multiple ECMP paths between two adjacent nodes, it can send two EPR Objects to the corresponding PCC, with the "Route Priority" field be set to the same value. Thanks, Mike.
_______________________________________________ Pce mailing list [email protected] https://www.ietf.org/mailman/listinfo/pce
