Re: [netmod] Last Call: draft-ietf-netmod-snmp-cfg-03 (20131220)

[Juergen Schoenwaelder <[e-mail]>]

On Mon, Jan 06, 2014 at 03:10:08PM -0500, ietfdbh wrote:
> 
> Will a relevant standard SNMP MIB implementation be compliant if the
> YANG module is used to do the configuration?

This is fairly abstract and general question. For me, an SNMP MIB
implementation will be compliant if it complies to the SNMP MIBs it
implements. Most SNMP agents I have seen read configurations from CLIs
and/or configuration files. Hence, I believe it is possible to have
other configuration interfaces. I also believe they are being used to
setup SNMP agents. (Frankly, I never configured VACM using the SNMP
tables - I always resorted to configuration files or CLIs. But then
this might have just been my lack of having access to smart software
to do this job for me. ;-)

> Is it possible to configure an SNMP MIB using the YANG module (in a
> compliant manner) that would result in the implementation not being
> compliant to the MIB standard, whereas using SNMP (in a compliant manner)
> would result in the implementation being compliant?
> Put more simply, can using the YANG *cause* an implementation of the
> relevant MIB to be non-compliant to the MIB standard?

We are trying hard to ensure that this interworks nicely. If there are
any things that may cause problems that we might have missed, by all
means let us know so that we can take a look at it. For example, we
might have to say something about spin locks to handle situations
where a sequence of SNMP set operations in interleaved with NETCONF
updates. This is very useful input and appreciated.

The goal of this effort is to provide implementors with a YANG SNMP
configuration interface that to the best of our knowledge does not
cause conflicts with the implementation of SNMP configuration MIB
modules and I believe the IETF is the best place to do this work. The
alternative of doing this work is to let every NETCONF implementor
come up with his/her own model that may but various undocumented
constraints on the SNMP interface - I do not think this helps with
SNMP interoperability (and it surely does not help with NETCONF
interoperability either).

So please, tell us about any technical issues we might have missed.
This helps us much more than abstract discussions.

/js

-- 
Juergen Schoenwaelder           Jacobs University Bremen gGmbH
Phone: [phone]         Campus Ring 1, 28759 Bremen, Germany
Fax:   [phone]         <http://www.jacobs-university.de/>