Hi Everyone,
Thank Joel, Yao, Andrew and Jeff very much for their valuable comments in
IETF 117 SPRING WG meeting.
My responses are inline below with [HC].
Best Regards.
Huaimo
C1: not sure there is a clear procedure to differentiate the two cases: whether
it is wrong or failed SID.
[HC]: The two cases: whether it is wrong or failed SID can be
differentiated in the following way:
IF there is a RIB/FIB entry for a SID (e.g., SID-N) and then
the entry for the SID is removed from RIB/FIB after a SPF,
THEN the SID (e.g., SID-N) is a failed SID.
I will add a procedure into the draft.
C2: There is a sentence in the draft "Distributing a BSID is out of scope". You
could make it more visible to the WG.
C3: It may be out of scope of the draft. But you cannot solve the problem
without it.
[HC]: I will add some texts to make it more visible to the WG.
C4: I disagree with the problem statement. You complicate the control plane and
data plane. The problem you are trying to solve is a corner case.
[HC]: The problem consists of two Fast ReRoute (FRR) protection
sub-problems:
1). Protect node SID of a node without binding SID
after IGP convergence and before a global backup path installed.
2). Protect node SID of a node with binding SID
(after IGP convergence and before a global backup path installed).
There is a SPRING WG document for resolving sub-problem 1).
This draft proposes a different solution, which may be simpler.
We may use binding SIDs, thus, it seems there is a need for protecting
binding SIDs.
_______________________________________________
spring mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/spring
