RE: Help/guidance on L2TPv3 MIB draft

["Natale, Robert C (Bob)" <bnatale@lucent.com>]

> -----Original Message-----
> From: Steve Moulton [mailto:moulton@snmp.com]
> 
> Perhaps AgentX could receive some attention to fix
> this shortcoming, rather than force MIB writers to
> use artificial constraints that are not a part
> of, and rather unnatural to, SNMP.  This kind of
> artificial constraint (Andy might say CLR) does 
> little other than impede SNMP (and SNMPv3) acceptance.

To repeat:

It has nothing to do with any "shortcoming" in AgentX --
the IETF standard extensible agent protocol supports
singly instanced MIBs just fine (RFC2741, 4.2.1, again).

It is however an "artificial constraint", of singly
instantiable MIBs.  It is that defect in many existing
MIBs (not its correction) that has, in part, retarded
SNMP deployment.

From an information engineering perspective, all MIBs
should be designed for multiple instantiability.  That
quality is (and always has been) readily available (with
trivially incremental design effort and runtime cost) in
the existing SMI -- and it would virtually have to be,
since multiple instantiability is a fundamental attribute
of shared data models.  Single instance runtime configs
are merely a natural subset.

Cheers,

BobN