CURRENT_MEETING_REPORT_

Reported by Michael Erlinger/The Aerospace Corporation

Minutes of the Remote LAN Monitoring Working Group (RMONMIB)


Monday

The RMONMIB Working Group met twice at the Toronto IETF. Due to some
scheduling conflicts for the formal presentations, the initial meeting
focused on general issues for the next version of RMON and a status
review of the current RMON draft.

The charter was discussed.  No changes were suggested, but the group did
note the short period in which a new RMON must be completed.  It was
recommended that the working group hold interim meetings focused on
various issues to be considered for a new RMON. Also formal proposals
would be solicited for new features.  Possible dates for an interim
meeting would be forthcoming.

The Monday meeting concluded with a discussion of the current RMON draft
(17 June 1994).  A Last Call was held within the working group and the
draft was forwarded to the area director with a recommendation that the
draft become a Draft Standard.  Members of the Network Management Area
Directorate have reviewed the draft and have suggested changes (textual
only).  As a result of the July interoperability testing, some issues
were raised.  For those issues that do actually pertain to RMON, textual
changes will be made at the direction of the Network Management Area
Directorate.


Tuesday

The Tuesday RMON session was devoted to presentations on various topics
for the new RMON. Copies of these presentations are available via
anonymous FTP from jarthur.cs.hmc.edu in the pub/rmon/toronto directory.

Nevil Brownlee presented an overview of the accounting MIB and its
architecture.  He has implemented the MIB on a number of platforms
(freely available implementation).  Rules are used to describe the
traffic patterns of interest.  Nevil has some data on performance issues
related to agent processing.

Sonia Panchen made a presentation of Statistical Sampling for network
traffic monitoring.  Hewlett-Packard has released a product, EASE, that
demonstrates a practical implementation of sampling for traffic
monitoring.

Venkat Rangan presented MIB structures to include network address
information in the RMON structure and to handle variable length headers
in the filter area.  These modifications were part of the RMON
extensions implemented by Hewlett-Packard.

Timon Sloane and Karl Auerbach presented possible MIB extensions to
allow for a packet generation capability in RMON and possible changes to
the existing filter mechanism.  Karl also described his stack machine
for filter processing.

Steve Waldbusser presented his list of pressing issues for a new RMON:
new filter language, network layer host and matrix, protocol type
distribution, trap destination, serial line configuration, issues of
static versus extensible and repeater/bridge ports.  Discussion
indicated little interest in serial line configuration, but the other
issues remain of interest to the group.


   o Filter language:
      -  needs to handle variable length fields
      -  simple to implement
      -  hopefully already defined
      -  efficient encoding

   o Berkeley packet filter:
      -  three to four generations of refinement
      -  implementations available
      -  space-efficient encoding