Re: BFD MIB extension for MPLS-TP

[Jeffrey Haas <[e-mail]>]

[Mostly writing as an individual contributor.]

On Thu, Dec 15, 2011 at 08:17:24PM -0800, Sam Aldrin wrote:
> We have submitted a new draft, BFD MIB extensions to support MPLS-TP, prior
> to IETF82 at Taipei.
> 
> The draft is located at
> http://datatracker.ietf.org/doc/draft-vkst-bfd-mpls-mib/
> 
> 
> We have presented the draft in MPLS WG to get feedback from that WG.
> 
> As there was no WG session at Taipei, we are seeking feedback from BFD WG
> over the mailing list.
> 
> 
> The MIB extension is fairly simple and adds few new objects to extend the
> exiting BFD MIB and support BFD sessions over TP tunnels. Please provide
> your feedback and comments, so that, we could take it forward and ask to
> make it a WG document.

My first comment is that the object names in this MIB seem to strongly
suggest that this is a general extension MIB for BFD rather than are adding
MPLS-TP specific objects to the BFD MIB.  E.g.:
bfdExtObjects
bfdSessExtTable
etc.

I would strongly suggest the object names be more specific in alluding to
MPLS-TP.

SessionMapTypeTC should similarly note that it is associated with both BFD
and MPLS-TP.  Also, if there's any chance that this TC will see re-use, it
should be placed in its own MIB at some point in the future.  IMO, it's fine
to leave it in this document for a few iterations for refining the MIB.

bfdSessExtTable contains read-create objects but is INDEXed on the BFD
session table rather than AUGMENTing it.  As such it needs its own RowStatus
and StorageType objects.

bfdSessExtMapType and bfdSessExtMapPointer should probably get useful DEFVAL
entries that install an inactive row if the MIB user inadvertantly do a
createAnd* operation.  This may suggest that a map type of "none" may be
required.

-- Jeff