Re: Adoption of draft-vkst-bfd-mpls-mib

[Jeffrey Haas <[e-mail]>]

Sam,

On Tue, Jun 05, 2012 at 05:35:37PM -0700, Sam Aldrin wrote:
> We were asked specifically by the WG chairs, in the Paris WG session, to add read write option to the MIB. AD and MPLS WG chair also provided their comments and felt BFD MIB extension to support TP should have write support.

This was under the presumption that we needed to provide this as part of our
MPLS-TP agreement with ITU.  After consulting with the MPLS chairs (or at
least, per Loa's answer) and our ADs, this doesn't appear to be the case. 

So, as a motivating factor, we don't need it.

> There are at least three vendors who are requesting write support as well.

*That* is far more interesting.  Per other discussions off-list, I was
considering polling the WG to see if we should consider removing the write
profile for the MIBs.

Would those three vendors be willing to go on record stating that they want
write-access to the MIB?  This wouldn't be a formal statement of "we'll
support this in X version of our software", just "we want the option to be
able to do SNMP SET in this MIB."  Having some sort of public statement for
this would help an upcoming OPS open-area meeting at the upcoming IETF in
Vancouver with regard to this general topic.

If the vendors aren't willing to publicly state this but are willing to
privately do so, that would also be helpful.

> To the original comment that there is no requirement for write support for MIB, my answer is, there is no requirement to have just read only either. If there is consensus that, write option shouldnt exist in MIBs, it should be stated so. Otherwise, the confusion will persist every time.

In general, a read-only profile is desirable because there exists a set of
vendors who have no intention to provide SNMP SET access to their MIB
implementations.  Without a supporting conformance statement, those vendors
are not conformant.  Obviously the protocol works just fine read-only, but
it upsets the people who track such things in spreadsheets. :-)

-- Jeff