CURRENT_MEETING_REPORT_


Reported by Bernhard Stockman/EBONE

Minutes of the Operational Statistics Working Group (OPSTAT)

Agenda


  o RFC1404 Deployment Experience
  o Client/Server Protocol:  Presentation by ANS
  o Statistical MIB


RFC1404 Deployment Experience

Daniel Karrenberg reported that the efforts to convert MONSTER to use
RFC1404 have not progressed.  Due to resource constraints further
development of MONSTER has been put on indefinite hold.  What would be
needed to make it use RFC1404 is just a relatively simple reformatter.
Further it could need a rewrite of the actual SNMP polling function
since this uses the snmp capable awk from ISODE which is very hard to
install.  Daniel renewed his offer to get started, those people who are
interested in continuing work on Monster.

Craig Haney who was not present at the Working Group meeting has a PERL
parser for the RFC1404 format.

A list of problems were reported with RFC1404:


  o In the syntax for device-fields multiple tag-tables are not
    delimited syntactically.  Proposed solutions:

     -  Count preceding
     -  Different field separator withing tag-table
     -  Bracketing tag-table with something BEGIN-TAGT, END-TAGT
     -  Bracketing with


The Group decided to use s around tag-table.  This will lead to a
revision of the BNF:


   device-section ::= "BEGIN_DEVICE" <FS>
                     <device-field> <FS>
                     "END_DEVICE"

   device-field   ::= <networkname><FS><routername><FS><linkname><FS>
                     <bw-value><FS><bw-sort><FS><proto-type><FS>
                     <proto-addr><FS><time-zone><FS> "{" <tag-table> "}"

                                  1





                     [ "{" <tag-table> "}" ]


Action:  Bernhard volunteered to check for similar problems and report
via mail.

After the discussion the Group decided to collect more problem reports
from implementors before trying to issue another RFC.

There IS interest in USING tools following the RFC out there.  There is
not much interest in actually producing the tools.  FARNET (who?)  might
produce some funding for developing PD tools.  Question:  Who would be
interested?  No immediate interest.

Action:  Bernhard will send a message to the list about this and try to
find capable implementors.  Anyone interested persons should contact
Bernhard.

Client/Server Protocol

The intention is to make (aggregated) statistical information available
from a server.  A strawman document is being developed.  The following
reports were given:


  o Henry Clark reported on implementation experience.  Login is
    primitive direct string comparison.  Separate configuration file
    defining the view of the users.  Server runs use psi formant but
    don't use opstat.

  o Ittai Herhman reported on a similar system currently being used at
    ANS which originated at Merit.  Refer to slides.

  o Vikas Aggarwal also showed some examples of reports.  Both Vikas
    and Ittai remarked that from limited experiments they have done it
    seems that for line utilisation there is little difference between
    15 minute and 1 minute sampling.  A preliminary document is being
    done.


The Group went into architectural brainstorm mode and reached the
following conclusions:


  o There is agreement to generalise the authorisation passed by login
    to generic authentication strings and leave in the authentication
    method used as a local matter beyond the scope.

  o Select shouldn't return the data, there should be a separate "get"
    command to initiate that, multiple selections are possible and
    identified by a selection ID which can be specified by client and
    will be returned by server.

                                  2





  o Status command to find out status of what is currently selected
    especially the amount of data.


Action:  Bernhard Stockman will change the strawman accordingly.

Statistical MIB

Consensus is that a short list of HIGLY DESIRABLE variables should be
agreed upon and forwarded to the ??  WG for inclusion in the next
Internet Standard MIB.

Action:  Bernhard will forward a list of all so far proposed variables
to the list and re-start discussion.

Attendees

Vikas Aggarwal           [email protected]
David Bolen              [email protected]
Erik-Jan Bos             [email protected]
Rebecca Bostwick         [email protected]
Douglas Carson           [email protected]
James Cassell            [email protected]
Henry Clark              [email protected]
David Conrad             [email protected]
Roger Fajman             [email protected]
Stefan Fassbender        [email protected]
Dale Finkelson           [email protected]
Eric Fleischman          [email protected]
Paul Franchois           [email protected]
Kenneth Goodwin          [email protected]
Ittai Hershman           [email protected]
Daniel Karrenberg        [email protected]
Kim Long                 [email protected]
David Lovering           [email protected]
Daniel McRobb            [email protected]
Greg Minshall            [email protected]
Bob Morgan               [email protected]
Marsha Perrott           [email protected]
Robert Reschly           [email protected]
Steven Richardson        [email protected]
Vilson Sarto             [email protected]
Richard Woundy           [email protected]



                                  3