From: (Keith Klamm)
Subject: Re: 802.1p/D3 Ballot
To: jlarson@fnal.gov
Date: Mon, 10 Jun 1996 11:17:40 -0500 (CDT)
Cc: p8021@hepnrc.hep.net


SUBJECT:  P802.1p/D3 - Traffic Class and Dynamic Multicast Filtering
          Services in Bridged LANs


__X__ I disapprove for the following reasons.


        Keith Klamm

----------------------------------------------------------------------
General Comments
================

I can not approve this standard until the issues of unicast registration
and priority queuing are resolved. These issues were discussed at the
.1 Interim meeting in Wakefield. I would like to see unicast registration
removed from this standard since this capability is better handled by a
unicast frame priority field. If .1 resolves to add priorities to
ethernet frames then .1p should remove all references to priority
queuing since this becomes an end-station designated capability and a
bridge vendor differentiating feature.

Technical Comments
==================

Comment 1:

pg. 13, line 38 - It should be made clear here that it is still
acceptable to build bridges that only support Basic Filtering
Services. Suggested change on line 38, change "on some Ports" to
"on some or all Ports".
         ^^^^^^

Comment 2:

pg. 26, line 35 - Since Static Filter entries can include individual
(unicast) addresses as well it should be so stated. Suggested change on
line 35, change "Static Filtering Entries shall include group addresses
and the broadcast address" to "Static Filtering Entries shall include
individual addresses, group addresses and the broadcast address"
^^^^^^^^^^^^^^^^^^^^^

Comment 3:

pg. 27, line 22 - This paragraph describes the use of a short timer
to age the FDB during topology changes. It would be helpful to reference
the paragraph (4.3.5) where the value of the short timer is defined.

Comment 4:

pg. 30, line 37 - I would hope that no protocol coming out of .1 would
require microsecond accuracy, this seems much to tight. See next
comment.

pg. 75, line 30 - Obviously this standard can't be approved without
timer specifications. I would recommend a timer granularity in the
range of 20 to 100 milliseconds.

Comment 5:

pg. 41, line 3 - Section 7 incomplete.

Comment 6

pg. 42, line 48 - "tergivesation" not in my dictionary.