Re: [Isis-wg] Proposed isis-wg documents - draft-ginsberg-isis-sbfd-discriminator-00

Hi Uma, Les,

Pardon me for jumping in.
Please see [NOBO] in-line.

> -----Original Message-----
> From: Isis-wg [mailto:isis-wg-bounces@ietf.org] On Behalf Of Uma
> Chunduri
> Sent: Thursday, August 21, 2014 3:00 AM
> To: Les Ginsberg (ginsberg); Christian Hopps; isis-wg@ietf.org list
> Cc: Hannes Gredler
> Subject: Re: [Isis-wg] Proposed isis-wg documents - draft-ginsberg-isis-sbfd-
> discriminator-00
> 
> Les,
> 
> In-line [Uma1]:
> 
> ===
> > Support this.
> >
> > Qs:
> >
> > 1. What happened to the discussion on-  if this discriminator is a
> > link property or node property as defined (asked originally in OSPF WG
> > by Hannes).
> 
> The question was answered - discriminators are a node property.
> 
> [Uma1]: I saw SR use cases for SBFD and with ADJ-SID's (link specific)  this
> can be a problem.

[NOBO] Couple of comments.
1. S-BFD document for Segment Routing is very stale. Authors should be able to roll out a revised and updated document this weekend.
2. S-BFD authors wanted to build in some connectivity test aspect into the S-BFD mechanism, particularly to allow verifying where Adj-SID terminates on the nextshop. However, there were strong push backs from the BFD WG regarding usage of BFD or S-BFD for connectivity test. Thus, S-BFD focuses only on continuity check, which is a node reachability check.

> 
> > 2. If MTs are on the same physical topology, I don't see the
> > association of Discriminator to the logical topology/MT.  Is it
> > already though and discounted ? If yes rationale will be helpful.
> 
> Just as IP/IPv6 addresses are NOT topology specific, S-BFD discriminators are
> NOT topology specific. The paths to reach a node may differ between
> topologies - but that is not a property of the discriminator.
> 
> [Uma1]:  A node may  be part of any  logical topology (MT) or may not be.
> But, if the discriminator doesn't represent the same then the receiving node
> (participating in multiple logical topologies) has no idea  for which topo the
> same has been received or In other words  sending node may not be part of
> the particular logical topo.
> 
> Do you see this as an issue?

[NOBO] This is a very good comment Uma. My response would be that there's nothing in S-BFD documents that prevents tying of local S-BFD discriminator to set of incoming interfaces. i.e. a local S-BFD discriminator for an MT can programmed to process S-BFD packets coming in only from set of interfaces which belong in the MT. If such strict checks are desired, implementations are free to do so.

Thanks!

-Nobo

> 
> --
> Uma C.
> 
> _______________________________________________
> Isis-wg mailing list
> Isis-wg@ietf.org
> https://www.ietf.org/mailman/listinfo/isis-wg