The following are the minutes from IETF-106 for BFD. Thanks to Xiao Min for
providing transcripts to help produce them.
Please send comments on the minutes ASAP.
-- Jeff
Bidirectional Forwarding Detection (bfd)
IETF-106, Singapore
Tuesday 17:10
Minutes taker, Xiao Min.
Chairs update - see slides.
- BFD for vxlan, presented later.
- BFD for large packets, presented later.
- BFD yang documents needed small change to accommodate BFD Unsolicited WGLC.
Reshad has sent update to RFC editor.
- BFD authentication documents. We have weak, but positive direction to send
those to IESG.
- Work deferred on BFDv2/extensions until next IETF. Seeking commentary from
IESG regarding charter discussion.
BFD for vxlan - presented by Greg Mirsky:
- Cover known open issues.
- Main discussion remains on target addresses and whether to use the 127/8
addresses always, or to permit implementations to deviate from that.
+ Interoperability issues are known when they're not consistent.
+ There are existing implementations that don't use 127/8.
+ A SHOULD is required here, along with discussion about not doing it this
way in the security considerations. (E.g. cut and paste from RFCs
4379/8029)
- Demand mode requires no additional text.
- Echo mode is out of scope.
- Once the above has been updated in the draft, we'll send to IESG.
BFD for large packets - presented by Jeff Haas:
- Presentation given, see slides.
- Additional operational considerations integrated.
- Need additional text covering S-BFD.
- Will send back to group for continuing WGLC after this has been done.
One-Armed BFD - presented by Weiqiang Cheng:
- Presentation given, see slides.
- Technology largely (completely?) overlaps TR-146 from BBF.
- TR-146 has a number of errors.
- Does BBF have IPR on this mechanism?
- Comparison to LSP self-ping
- Matthew Bocci points that we don't have any rate negotiation we can do here.
+ Weiqiang responds that for low rate, we can verify connectivity, and maybe
low-grade link quality metrics.
- Greg Mirsky rightly points out this only works single-hop, and that contents
of echo packets may need to be discussed. Exactly what can we measure using
this technique?
- Robert Raszuk (jabber) asks about spoofing.
+ Greg had noted single hop.
+ Jeff: Sender would have to ensure packet contents provide security.
- We will need to followup on this on the mailing list.
