home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Internet Info 1997 December
/
Internet_Info_CD-ROM_Walnut_Creek_December_1997.iso
/
ietf
/
acct
/
acct-minutes-92mar.txt
< prev
next >
Wrap
Text File
|
1993-02-17
|
3KB
|
89 lines
This is only a rough draft - Megan 04/08/92
Internet Accounting Working Group Meeting Minutes
17-Mar-92 to 18-Mar-92
Internet Accounting Architecture
The Tuesday session began with a presentation by Jim Alfieri of
Bellcore. Jim spoke on Bellcore' plans for SMDS (Switched Multi-megabit
Data Service) usage accounting. He mentioned that packet counting was
done upon egress from the switch (in our last session, the issue of
where to count was raised and the WG agreed that the architecture should
support both counting on entrance and counting on exit).
Jim explained that the SMDS accounting service does not use the Meter
-> collector -> application model, where the collector aggregates
frequent meter reports into a usage file. In the SMDS case, the meter
(inside the SMDS node) itself does the aggregation of usage data to a
file. This file, in AMA TPS (Automatic Message Accounting
TeleProcessing System) format, is periodically transferred directly to
the billing system application via FTP. A drawback of this architecture
is that FTP service provides no security. Further details can be found
in Bellcore Technical Advisory TA-TSV-001062, "Generic Requirements for
SMDS Customer Network Management Service".
The rest of the session focussed on the most recent version of the
Internet Accounting Architecture document. Working group participants
offered comments, criticisms and suggestions that will be incorporated
into the next draft. In particular the following modifications will be
made.
o Explain that it is the collector's responsibility to manage
time and determine/interpret the meaning of meter ticks.
o Discuss space reclamation strategies for dealing with memory
exhaustion in the meter.
o Discuss the problem of trap storms that may occur when the
meter is trying to avoid losing data due to memory exhaustion.
o Discuss the possibility of trapping to other/multiple managers.
o Discuss the (in)applicability of the RMON MIB.
We believe that when the WG input has been incorporated in the document
and it has been reviewed by the mailing list it will be ready to become
an Internet Draft.
Internet Accounting MIB
On Wednesday the WG discussed the first real draft of the IA MIB.
Many detailed (but important) changes were suggested and will be
incorporated into the next version of the architecture document. Among
them:
o Put a general explanatory section up front.
o Add/improve the descriptions of special values for
parameters. As a special case, zero values should be used in
certain cases (e.g. high water mark or flood mark) to disable
a feature.
o We need to add text explaining flows.
o The MIB's "address type" should be expanded to include (a)
layer number and (b) type (e.g. if layer = 2, allow all
interface designations from the standard MIB).
o Use standard SNMP timeticks (1/100 second).
o Explain what max lifetime is good for.
o Recommend UDP checksums.
The WG will conduct a discussion of these changes and review this
document over the Internet in the next couple of months with a view of
advancing it to the status of Internet Draft by the next IETF
conference.
General
The WG discussed what should be done when the architecture and MIB
documents are completed. It was generally felt that implementation of a
prototype accounting system was next. This could be done either through
a revision/extension of the WG's charter, or by forming a new,
successor WG.