Minutes for SNA NAU MIB WG

[Deirdre Kostick <[e-mail]>]

Folks --

Attached are the minutes from
the 7/26+/94 meeting of the 
SNA NAU MIB working group


ALSO --

Note that the SNA NAU MIB is
now RFC1666. 

Deirdre
===============minutes=============================



SNA NAU MIB WG - Meeting 7/26/94

Chair(s):
	Zbigniew Kielczewski <[e-mail]>
	Deirdre Kostick <[e-mail]>

Network Management Area Director(s) 
    Marshall Rose <[e-mail]>
 
Mailing Lists:
   General Discussion: [e-mail]
   To Subscribe:      [e-mail]
   Archive:           thumper.bellcore.com:pub/tob/snanaumib

Meeting Minutes SNA NAU MIB WG Meeting - 7/26/94

The SNA NAU MIB working group met on July 29, 1994 
to review the proposed MIB objects for APPC 
<draft-ietf-snanau-appcmib-00.txt>

APPC and APPC MIB Overviews
===========================
Mike Allen presented an overview of APPC.
Bill Kwan presented an overview of the current
APPC MIB structure. 

APPC MIB I-d Review
======================== 

The working group walked through the current
draft and identified the issues/recommendations
summarized below. MAILING LIST INPUT ON
THESE RECOMMENDATIONS/ISSUES IS ENCOURAGED.

Issues/Recommendations Summary:

o Need to define traps for SNMP-based management of APPC.
The options identified for defining traps for the APPC MIB
are:
  + translate all the  existing APPC-specific alerts into traps
  + use a subset of the alerts 
  + don't define traps

Actions Items: Bill Kwan will post information on current
APPC-specific alerts to the mailing list.

o What information on conversations should be modelled in the MIB?
Should the MIB contain information only on active, historical, or both
conversations?

Concerns were the amount of information that an agent would be required
to keep, giving some guidance to implementors on the relative size
of the tables. One alternative discussed was to have two tables:
one table containing a log of previous conversations with a separate
"trace" table containing a limited set of information on current
conversations. 

Action Items: 

Mike Allen and Richard Daugherty will investigate
three options for modelling sessions and conversations:
   o Active conversations [sessions] with information on the
     previous 1 or 2 state changes
   o Active conversations with  a finite size table, or
   o Active conversations with a variable-size table.
They'll post a recommendation basically for what / how much
info should be kept. 

Mike Allen will post info on what NetView supports today
for conversations (to give the working group a sense of
the type of management information and capabilities they
have today).

Howard Berkowitz will post relevant example from the RMON MIB
for control/trace functions. (DONE)

o appcTpOperTable

The WG recommended DROPPING the table since we cannot reliably
expect to have implementations that support listing all
the active TPs on a node.

Related change:
The appcTpOperIndex will be dropped from the conversation table.

o appcGlobalTable

The WG recommended DROPPING the appcGlobalTable. The reason
is that no specific usage could be identified for the table.
The SNA NAU mib (RFC1665) contains information on LUs and
sessions per node, although the information presented in
the SNA NAU mib is not summary info. The information in
the appcGlobalTable would not give an indication of
traffic or load.

o APPC Statistics Control

Add an object to control conversation tracing on/off.
(This assumes that a trace table would be implemented
for conversations.)

REMOVE appcStatCntrlCtrStatus and appcStatCntrlCtrStatusTime
since these objects reflect implementation specific
features.

ADD a text indicating that when collection is turned off,
the counters will be set to zero.

The WG discussed issues related to multiple manager entities
controlling agent functions (such as trace on/off). The WG
decided that in the initial APPC MIB, the functionality would
be kept very simple, with the MIB containing only one 
"on/off switch". If implementation experience shows that
multiple manager scenarios cause problems that need to
be fixed by adding some control scheme to the mib, this
would be future APPC MIB feature.



o appcDefaultTpOperation object

The enumerated values reflect only alternatives for
one implementation; other implementations support other
values. QUESTION FOR MAILING LIST:

What other values should be added to the enumerated
list?

o appclLUOperDefType

The meaning/intended use for the DefType values (sys default & operator)
were not clear. WG recommended dropping this OBJECT, plus
other defType objects in the current version.

o appclLUOperTable

Need to add an Admin Table for the configurable objects
(operSessLength, BindResponseMaxQ)

o appcDefaultLuAlias

Change to appcDefault LuName. Remove appcLLUOperDefaultLu
since info is duplicated.

o FUTURE ITEM FOR RFC1665

Need to define additional values for LU ADMIN TERM.
There are other variations of UNBIND. 

o lLUOperTable

Need to add compression info.


o appcTpAdminPfCnos & appcTpAdminPfSessCntl 

REMOVE since these objects reflect implementation-specific
information.

o appcTpAdminConvType

SPLIT into two objects, reflecting
basic/mapped, fdx/hdx info

o appcTpAdminSecLvl

REMOVE since these objects reflect implementation-specific
information.

o appcTpOperTable

REMOVE

o appcSessnS2P/P2S* objects

Are these objects useful?
Counts of functional management headers (FMH) in
each direction does not seem useful; however,
FMH5 headers does seem useful.

Action Item:


o appcCPICAdminSecType

Need to add an object reflecting the security characteristics.

o CPIC Table


Discussion on the need for admin/oper tables.


 

NEXT STEPS
==========

o 8/15/94 - 
A revised working draft will be posted to the
mailing list. The revised draft will contain the changes
discussed during the meeting, except for the conversation
table updates, and revised oper/admin text. See the next
items. (Bill Kwan)

Michael Allen/Richard Dougherty will meet to review
alternatives for the conversation table/objects.
Mike will post their input to the list.

o 8/19/94 -

  + First cut at more descriptive text for operational/admin
object relationships. (Zbigniew)

  + APPC-specific alerts posted to list. (Bill Kwan)

  + Link Station objects posted to list (Michael Allen)

o 9/~20/94

  + Next Internet-draft

o 10/?/94
  + Interim meeting AIW

o 10/31/94

  + Internet-Draft


o PER THE SNA NAU WG SCHEDULE THE APPC MIB will be completed
by 10/94. There are currently NO PLANS to meet at the
next IETF meeting.

o DO WE NEED AN INTERIM MEETING??


Thanks to all who participated!!
Thanks to Bill Kwan, Zbigniew and Mike Allen
for preparing the APPC MIB proposal!

Deirdre

WORKING GROUP MEMBERS ARE
ENCOURAGED TO MAKE COMMENTS
ON THE MAILING LIST PREFERABLY
WITH DETAILED ADDITIONS,
WORDING, CHANGES, etc.