Re: MIB-II on proxied chassis entities.

kzm@hls.com (Keith McCloghrie) Tue, 21 July 1992 06:36 UTC

Return-Path: <owner-chassismib>
Received: by CS.UTK.EDU (5.61++/2.8s-UTK) id AA26635; Tue, 21 Jul 92 02:36:26 -0400
Received: from LANSLIDE.HLS.COM by CS.UTK.EDU with SMTP (5.61++/2.8s-UTK) id AA26631; Tue, 21 Jul 92 02:36:23 -0400
Received: from nms.netman (nms.hls.com) by lanslide.hls.com (4.1/SMI-4.0) id AA25145; Mon, 20 Jul 92 23:36:45 PDT
Received: by nms.netman (4.1/SMI-4.1) id AA02539; Mon, 20 Jul 92 23:26:13 PDT
From: kzm@hls.com (Keith McCloghrie)
Message-Id: <9207210626.AA02539@nms.netman>
Subject: Re: MIB-II on proxied chassis entities.
To: gallagher@quiver.enet.dec.com
Date: Mon, 20 Jul 92 23:26:12 PDT
Cc: chassismib@cs.utk.edu
In-Reply-To: <9207201944.AA12101@us1rmc.bb.dec.com>; from "gallagher@quiver.enet.dec.com" at Jul 20, 92 3:47 pm
Organization: Hughes LAN Systems
X-Mailer: ELM [version 2.2 PL0]


Shawn,
 
> >Of course, if there isn't a real agent after all, one *might* argue that
> >lineCard1's view doesn't need to contain a snmp group.
> 
> Okay, then I need to back up and ask a much simpler question.  Is my 
> assumption, "...the minimal MIB-II implementation has the system and 
> snmp groups" bogus?  
 
No, your assumption isn't bogus, but consider that for a regular (i.e. not
proxy) agent, each MIB view is a (proper or improper) subset of that agent's
MIB which (would normally but) need not include the system or snmp groups.

While there are proposals (e.g. SMP) to increase the formalism of
specifying conformance, it will (probably, can) never be specified to
the last degree, i.e., there will always be gray areas.  One of the
strengths of TCP/IP is that the market decides what is necessary, rather
than conformance spec.s (cf. the Host Requirements RFCs, to which very
few, if any, implementations fully conform.)

> I'll argue that having individual snmp groups in this case adds no value.
 
In my opinion, this is the true test of whether such groups are needed.
Note that if the snmp group were a table instanced by party, then it
would be different.

> Do you see the chassis MIB providing guidance in this area, 

No.

> or will it be left an implementation specific issue?

I would say implementor's choice.

Keith.