Re: Information Distribution

Bob Stewart <rlstewart@eng.xyplex.com> Thu, 24 September 1992 20:39 UTC

Return-Path: <owner-chassismib>
Received: by CS.UTK.EDU (5.61++/2.8s-UTK) id AA07993; Thu, 24 Sep 92 16:39:09 -0400
Received: from xap.xyplex.com by CS.UTK.EDU with SMTP (5.61++/2.8s-UTK) id AA07989; Thu, 24 Sep 92 16:39:04 -0400
Received: by xap.xyplex.com id <AA20569@xap.xyplex.com>; Thu, 24 Sep 92 16:37:38 -0500
Date: Thu, 24 Sep 1992 16:37:38 -0500
Message-Id: <9209242137.AA20569@xap.xyplex.com>
From: Bob Stewart <rlstewart@eng.xyplex.com>
To: chassismib@cs.utk.edu
In-Reply-To: SAPICH's message of Thu, 24 Sep 92 13:01:58 CET <9209241301.A05403@ccmailgate.conware.de>
Subject: Re: Information Distribution


>        1. How will the information of a new card (just plugged in)
>           be distributed in the system?
>           Are there any ideas of a common mechanism?

The mechanism is implementation dependent.  I can now, finally, say publicly
that we have new chassis with an enterprise chassis MIB.  We collect its
information over an out-of-band management bus, which seems to me like the
best way.  New or changed cards come and go dynamically.  Other mechanisms
discussed on the list include an operator-entered configuration file or
in-band communication with SNMP or some protocol of your own design.  It's all
up to what your hardware and software can do and how much having a Chassis MIB
is worth to you.  For our old chassis, without the management bus, we probably
won't implement any sort of Chassis MIB as it doesn't seem worth it.

>        2. Is there only 1 master for the chassis mib in the system
>           or can there be more masters? If there are more masters
>           how are they supervised?

There is no "master", in the sense of a point of control.  You can have one or
many agents that implement the Chassis MIB for a single physical or logical
chassis.  Some people prefer one, some prefer multiple.  We allow multiple,
but don't require any.  It depends on what you install in the chassis and what
you want.  The consensus we do have is that all instances of the Chassis MIB
for the same chassis should supply the same information.

	Bob