On March 20, 2019 at 7:37:42 PM, Les Ginsberg (ginsberg) ([email protected]) wrote:
Les: Hi! In regards to the relationship with RFC 7794, there is a history here. Thanks for the history lesson! :-) To avoid more confusion (specially from other reviewers), I think that it would be a good idea to avoid talking about transitions…. The logic of preferring the prefix attributes sub-TLV seems enough. When “someday” comes, and if it’s needed, we can worry about repurposing the flags. Thanks! Alvaro. The N and R flags were originally defined as part of the prefix-sid sub-TLV in early versions of this draft. However, it was realized that there were use cases for these flags that had nothing to do with SR. This prompted writing what became RFC 7794. But because there are early deployments in the field which were already using the N/R flags in prefix-SID sub-TLV we could not immediately deprecate the flags in prefix-SID sub-TLV. The intent is for RFC 7794 advertisement to become the sole place where N/R flags are advertised - but in the interim implementations have to support logic which is roughly: If (prefix attributes sub-TLV is present) Use prefix attributes sub-TLV settings and ignore N/R flag in prefix-SID sub-TLV if present Else Use N/R flag in prefix-SID sub-TLV if present "Someday" all implementations will support prefix attributes sub-TLV and there will be no use case for the flags in prefix-SID sub-TLV. At that point the flags can be repurposed. When will "someday" happen?? I think you know as well as I that such transitions take many years. The timing of this is not critical unless we run out of flag bits in the prefix-SID sub-TLV and need to reuse the N/R bits. At this point we do not foresee this happening any time soon.
_______________________________________________ Lsr mailing list [email protected] https://www.ietf.org/mailman/listinfo/lsr
