CURRENT_MEETING_REPORT_ Reported by Henry Clark/OARnet Minutes of the Operational Statistics Working Group (OPSTAT) The session stared with a review of the group's charter and history. The group's original and current charter is to look at operational statistics and methods of using and viewing them in operational networks. RFC 1404 is an intermediate result in developing tools and additional methods of allowing the exchange and viewing of data between networks or a provider and its customers. Client-Server After making a brief overview, Henry Clark discussed his work on the client-server data retrieval method. His server is complete, and the client is partially done. Several questions were raised: o Access control? A paragraph will be placed in the Internet-Draft about the need for access control through the server, but no specification of such will be placed in the document. o Data presentation to user? We discussed recommendations about client ``standard'' presentation methods. We agreed that since the data from the server should be returned in RFC 1404 format for portability, it may be that the client would offer a variety of formats to the user, ranging from saving in a file, to sending to a printer, to drawing a nice X-windows graphs. A later document of the working group may recommend standard report formats. o How aggregate vars to user? This was a meta-discussion on how individual SNMP variables might be aggregated into a single variable. This was seen as useful but extremely hard. This is a topic for future discussion. o How much of client in RFC? A paragraph making suggestions about some of the ways a client might present data to the user will be included, but no ``oughts'' or ``musts'' will be included. Some discussion of presentation versus storage started. Some folks, for efficiency reasons, may not store data in RFC 1404 format, but would run the server so that the data could be retrieved in RFC 1404 format. Accounting Stuff A host attached to a network observes data packets and builds a table of data flows. Flows the meter are interested in is described by a set of rules that match a pattern. Several graphs were displayed. If you are interested in this topic, go to the ACCT2 BOF on Wednesday. PREPnet/PSC PREPnet has been funded by NSF to query SNMP variables and store them in RFC 1404 format. Code should be available by the July IETF. They are using the CMU SNMP for queries. Some minimal presentation tools will be included. They hope to be done with the implementation by February 1995, and there is some money for maintenance. Suggested changes to RFC 1404 by Nevil Brownlee: 1. Change ``Kbps'' to ``kbps'' 2. Disallow white space as field separators; use commas within record and semicolon at end 3. Allow a single device section -- felt that was useful so make big files and not lots of little files 4. Multiple label sections: ptrs to files 5. Require brackets around tag table, variable list, tag name, and data value list Items 2, 4, and 5 were felt to be useful. Item 3, although seen as useful, would reduce functionality since small numbers of large files are useful in some environments. No opinions were expressed about item 1. Further discussion will be held on the mailing list. Goals and Milestones This section will be updated as: o Revised RFC 1404 Internet-Draft - July 94 o New Internet-Draft C-S July 1994 o Preliminary statistics document - November 94 Previously there had been some work surrounding a ``Statistics MIB.'' This issue may be revisited later. Attendees Jon Boone boone@psc.edu Michael Brescia brescia@bbn.com J. Nevil Brownlee nevil@ccu1.aukuni.ac.nz David Chen dchen@vnet.ibm.com Henry Clark henryc@oar.net Mark Knopper mak@aads.net J. Scott Marcus smarcus@bbn.com Ken Masica masic@es.net Daniel McRobb dwm@noc.ans.net Marsha Perrott perrott@prep.net Peter Phillips pphillip@cs.ubc.ca Bernhard Stockman boss@ebone.net