Re: SNMP object terminology

["Tom Petch" <nwnetworks@dial.pipex.com>]

Thanks for that; yes, it clarifies things.

Trying to be precise, I cited the wrong RFC.  I agree RFC 3411 is an enhancement
to RFC 1157; what I meant to refer to was section 3 of RFC1155.  Some of the
older RFC have a simple and clear statement of a complex subject and I find
myself referring people to them as a starting point; this section is one of
them.

Tom Petch

----- Original Message -----
From: "Keith McCloghrie" <kzm@cisco.com>
To: <nwnetworks@dial.pipex.com>
Cc: "mibs" <ietfmibs@ops.ietf.org>
Sent: Thursday, February 17, 2005 4:44 PM
Subject: Re: SNMP object terminology


> > Lately, I have been trying to be precise about the terminology used for SNMP
> > objects and after reading RFC 257x and RFC 341x have arrived at two
thoughts.
> >
> > First, there is a shift between SNMPv1 and SNMPv3, where the former spoke of
> > object types, object type instances or object instances and the latter talks
of
> > managed objects, types of managed objects and instances of managed objects.
Is
> > this a conscious shift, chosen to reflect some change in semantics?
>
> It's not a conscious shift, and reflects no change in semantics.
> I suspect it occurred because the word "object" is used (and possibly
> mis-used) much more often today, and thus "managed object" is an attempt
> to be eliminate any confusion with extra baggage that the word "object"
> has picked-up over the last 10+ years.
>
> > Second, this terminology does not seem to be spelt out anywhere in the
SNMPv3
> > library, in the way it is in RFC 1157 section 3; rather the terms crop up in
the
> > descriptions of indexing, of contexts, of security without ever being given
a
> > clear explanation.
>
> I think there are clear explanations but since SNMPv3 consists of more
> documents which are more modularised, the explanations are spread out
> across multiple documents and harder to find.  For example, I think
> section 3.3 of RFC 3411 is a more complete explanation of (managed)
> object types, instances and naming than RFC 1157 has; whether it is
> more/less clear is of course subjective.
>
> Keith.