home *** CD-ROM | disk | FTP | other *** search
-
- CURRENT_MEETING_REPORT_
-
- Reported by Cyndi Mills/BBN
-
- ACCT Minutes
-
-
- o Status Review
-
- - Internet Accounting Background Document
- - Internet Accounting Architecture Document
- * Reporting Format
- * Rule Table
- * Meter Control
- - Liason with other Activities
- * SMDS
- * RLANMON MIB/OPSTATS
- * Interop BOF
- * SNMP/Security
-
- o Working Agenda
-
- - Consensus on Internet Accounting Background Document
- - Consensus on Internet Accounting Architecture (Reporting
- Format)
- * Work on Rule Table
- - SNMP Concerns
- * Discuss Sample MIB
- * Find a home for the MIB
- * Security Concerns
-
-
- Internet Accounting Background Document
-
- The Working Group agreed to recommend the advance of the document to
- informational RFC when the following corrections are made:
-
- The language about security requirements needs to be made more precise.
- The security area will provide a review of the amended text. Mention
- that the collection protocol is responsible for providing the necessary
- security, and therefore details of the security mechanisms are outside
- the scope of the document.
-
- An explanation of the trade-offs between accounting on entry and exit
- (or both) in a router should be included. This discussion should
- include examples of performance impact, billing for offered load vs.
- billing for delivered load, reconciliation of counts with neighbors, and
- fragmentation. Note that SMDS has chosen to count on exit only.
-
-
-
- 1
-
-
-
-
-
- Internet Accounting Architecture
-
- The Accounting Architecture Document will be edited and then placed in
- the Internet-Draft area for access before the next meeting. The
- Architecture Document currently covers much of the material which was
- originally intended for the Meter Services Document. Mark Seger
- contributed many of the ideas and deserves special mention.
-
- In the reporting format, the current link-level address is an
- insufficient description. It is amended to be the ``adjacent address'',
- as in the address of the next lower protocol layer carrying the internet
- packet. This concept should be explained in the architecture. OSPF
- describes a similar adjacency and may provide a useful explanation.
-
- The motivation for the polling vs. interval reporting discussion in the
- architecture document should be made clearer.
-
- The rule table was reviewed and will be re-issued in another format. It
- was agreed that some form of rule pre-processor will probably be needed
- for network manager sanity.
-
- The manner in which the rule table for internet accounting forms a tree
- structure should be related to routing trees.
-
- The notation for the rule table should be amended. The binary pattern
- matching scheme should be made up of the character set ``0 1 *'', where
- ``0'' requires that the corresponding bit in the value to be searched be
- 0 for a match to occur, ``1'' that the bit be 1, and ``*'' where then
- corresponding bit may be either 1 or 0. The overall notation should be
- restructured to make the formation of a MIB easier.
-
- Because of the limitations of SNMP, it is difficult to access accounting
- information in table format. The appropriate set of protocol
- contortions needs to be investigated to return accounting information
- accurately and efficiently.
-
- The maximum lifetime of a flow should be determined by the managing
- entity rather than by the meter. The manager must be able to maintain a
- clean state - e.g., insure that a record is fetched and a new record is
- started for the same flow as an atomic operation. Also, the fetched
- record should be stored at least temporarily (for a ``short'' time,
- ``short'' to be determined by the manager) in order to allow for a
- retransmission (i.e., repeated request).
-
- The mechanisms for controlling data loss should be simplified to one or
- two parameters. If a meter is in danger of buffer overflow, probably
- the buffers will overflow and data will be dropped before the manager
- can take action anyway. This need further consideration.
-
- The flags grouping data need further definition. The notion that some
- group of flows may be categorized as ``expendable'' (discard these flows
- first) or ``essential'' (avoid discarding these flows if at all
- possible) should be further explained.
-
-
- 2
-
-
-
-
-
- SNMP Concerns
-
- Jesse Walker wrote a sample MIB illustrating techniques that will be
- needed for expressing the accounting reporting format in terms that are
- compatible with SNMP.
-
- The possibility of including of the MIB as a subtree of the RLANMON MIB
- was discussed and rejected by the Working Group. The two chief reasons
- for this were:
-
-
- 1. The RLANMON MIB is currently progressing towards draft standard and
- the late addition of accounting might hinder their progress.
-
- 2. Devices other that remote LAN monitoring equipment also perform
- accounting functions, so it is inappropriate to demand that these
- other devices implement the remote LAN monitoring MIB. It was
- pointed out that the RLANMON MIB could be structured so that only
- the accounting option is present in the MIB, but this was not
- considered a sufficient reason for homing the accounting MIB to
- RLANMON.
-
-
- Internet Accounting requires that SNMP be able to dump tables in a more
- efficient and atomic fashion. FTP might be a preferred method of
- reporting data, but not all meters can be expected to support FTP due to
- memory limitations and abbreviated protocol stacks.
-
- There are concerns that SNMP security may not be sufficient for
- accounting purposes.
-
- Note change in location of list.
-
- To join the accounting Working Group list, please send mail to
- accounting-wg-request@wugate.wustl.edu with ``SUBSCRIBE'' in the subject
- line. To leave the accounting Working Group mailing list, send mail to
- the same address with the subject line ``UNSUBSCRIBE''.
-
- Attendees
-
- Gigi Chu gigic@hpspd.spd.hp.com
- Tracy Cox tacox@sabre.bellcore.com
- James Davin jrd@ptt.lcs.mit.edu
- Alan Emtage bajan@cc.mcgill.ca
- Shawn Gallagher gallagher@quiver.enet.dec.com
- Phillip Gross pgross@nis.ans.net
- Ittai Hershman ittai@nis.ans.net
- Mark Hoerth mark_hoerth@hp0400.desk.hp.com
- Mike Janson mjanson@mot.com
- Deidre Kostick dck2@sabre.bellcore.com
- Tim Lee-Thorp ngc!tim@uunet.uu.net
- Peter Liebscher plieb@sura.net
- Joshua Littlefield josh@cayman.com
-
-
- 3
-
-
-
-
-
- Cyndi Mills cmills@bbn.com
- Dennis Morris morrisd@imo-uvax.dca.mil
- Bradford Parker brad@cayman.com
- Robert Reschly reschly@brl.mil
- Ron Roberts roberts@jessica.stanford.edu
- Kary Robertson kr@concord.com.kr
- Gregory Ruth gruth@bbn.com
- John Scudder jgs@merit.edu
- Robert Shirey shirey@mitre.org
- Anil Singhal
- Roxanne Streeter streeter@nsipo.nasa.gov
- Ronald Tencati tencati@nssdca.gsfc.nasa.gov
- Claudio Topolcic topolcic@nri.reston.va.us
- William Versteeg bvs@nrc.com
- Jesse Walker walker@eider.enet@decpa.dec.com
- Gerard White ger@concord.com
- Osmund de Souza desouza@osdpc.ho.att.com
-
-
-
- 4
-