Re: Chassis MIB comments

dan@lannet.com (Dan Romascanu) Tue, 25 August 1992 04:32 UTC

Return-Path: <owner-chassismib>
Received: by CS.UTK.EDU (5.61++/2.8s-UTK) id AA19729; Tue, 25 Aug 92 00:32:35 -0400
Received: from [192.84.3.7] by CS.UTK.EDU with SMTP (5.61++/2.8s-UTK) id AA19716; Tue, 25 Aug 92 00:31:55 -0400
Received: from moon.lannet.com ([149.49.50.12]) by lannet.com (4.1/3.1.090690-Lannet Data Communications) id AA05336; Mon, 24 Aug 92 09:52:50 IDT
Received: by moon.lannet.com (4.1/SMI-4.1) id AA22840; Mon, 24 Aug 92 09:52:48 IDT
Date: Mon, 24 Aug 92 09:52:48 IDT
Message-Id: <9208240652.AA22840@moon.lannet.com>
From: dan@lannet.com (Dan Romascanu)
To: kzm@hls.com (Keith McCloghrie), chassismib@cs.utk.edu
Subject: Re: Chassis MIB comments


>> Concentrator: A (device implementing a) node that has additional ports
>> beyond those required for its own attachement to the network. (Ref.: Kessler,
>> Train - MAN - Concepts, Standards and Services).

> This seems to me a very loose definition.  To the extent that it applies
> to concentrators, it seems to apply equally well to repeaters, bridges, 
> or even routers - without redundant paths, disabling any port loses
> connectivity from "its own attachment" to somewhere on the network.

You are right, but this is exactly the intention. 'Concentrator' in my
view applies also to repeaters, bridges, routers as long as they implement
internally the node topology in the definition.
 
>> Repeater: Device that extends the geographical coverage of a network by
>> interconnecting two similar LANs, such as Ethernet or Token Ring. Operating
>> at the physical layer of the OSI layer, it repeats (amplifies, reshapes,
>> retimes) packets received from one LAN before sending them to the other.
>> (Ref.: Terplan - Communications Network Management). 

> The use of "physical layer" excludes bridges and routers, but why are
> concentrators excluded ?  Shouldn't it say bit-wise, rather than "packets",
> and "one LAN-segment" rather than "one LAN" ?

The definitions do not exclude one another (this is the reason we use
a sum and not an enumerated INTEGER, isn't it?).
'Concentrators' and 'Repeaters' definitions are overleaved. Repeaters are
probably always concentrators as well, but concentrators may be passive
devices (which do not amplify, reshape, retime) which are not repeaters.
This second case is not covered by the definition in its present form.

I do not intent to trigger a phylosophical discussion on this subject, which
is far from being the most important in the Chassis MIB. However, my point is
that in the first draft we had 'repeaters' and 'concentrator'- no 'MAU', now
we have 'repeater' and 'MAU'- no 'concentrator'. In order to be consistent we
should either include all the three or define just one 'physicalLayerDevice'.

dan@lannet.com

Dan Romascanu
Systems Group Manager

LANNET Data Communications
Tel Aviv, Israel

Voice: 972-3-6458414
Mail:  972-3-5447146