OPS Area
RMONMIB WG Meeting Minutes
47th IETF 
Adelaide, Australia
March 29 and 30, 2000
Minutes by Andy Bierman

Review Material
---------------

(A) draft-kzm-hcdata-types-05.txt
(B) draft-ietf-rmonmib-hcrmon-06.txt
(C) draft-ietf-rmonmib-rmon2hc-00.txt
(D) draft-ietf-rmonmib-rmonhc-00.txt
(E) draft-ietf-rmonmib-rmonprot-ref-01.txt
(F) draft-ietf-rmonmib-rmonprot-mac-02.txt
(G) draft-ietf-rmonmib-iftopn-mib-00.txt
(H) draft-ietf-rmonmib-dsmon-mib-00.txt
(I) draft-warth-rmon2-artmib-01.txt
(J) draft-dietz-tpm-mib-00.txt
(K) draft-waldbusser-rmonmib-apm-00.txt
(L) draft-bierman-rmonmib-apmcaps-00.txt
(M) draft-steinberger-mib-vircir-00.txt
(N) draft-ietf-rmonmib-rmonfull-02.txt

Agenda
------

1) Status of Internet Drafts under IESG review and 
   the HC-DATA-TC MIB, which affects the HC-RMON MIB 
   (A - F, N)

2) Mailing List Issues
   - MIB issues raised on rmonmib@cisco.com since the
     last WG meeting 

3) Discussion of the IF-TOPN MIB (G)

4) Discussion of the DS-MON MIB (H)

5) Discussion of Virtual Data Source Issues

6) Application Performance Measurement (I - L)

Minutes
-------

1) WG Status 

The WG Chair presented a status summary of the I-Ds
before the WG.

Progress on the HC-RMON MIB (B) and the related 
RMON MIBs with HC TopN support (C and D), has been 
held up due to the 2 Counter64 based TCs in the HC-RMON MIB.

The HC-DATA I-D (A) was published to address this issue,
and that I-D has just been approved for RFC publication
by the IESG.  

The HC-RMON MIB will be republished with the following changes:
  - RmonGauge64 TC removed
  - ZeroBasedCounter64 TC removed
  - CounterBasedGauge64 and ZeroBasedCounter64 IMPORTed
    from HCNUM-TC
  - usage of RmonGauge64 will be replaced with CounterBasedGauge64

The RMON-1 MIB in SMIv2 format (N) has been approved by the IESG
and is awaiting publication as a Full Standard RFC.

The RMON Protocol Identifier Reference (E) has been received by 
the Area Director, and is awaiting Peer Review. 
The WG has requested that this draft be published as a
Proposed Standard RFC.

The RMON Protocol Identifier Macros document (F) has been received
by the Area Director,  and is awaiting Peer Review. 
The WG has requested that this draft be published as an
Informational RFC.

The WG Editor agreed to update the Token Ring RMON MIB (RFC 1513)
in SMIv2 format.  No progress has been made on this task.

2) Mailing list issues

2.1) HC-RMON mediaIndependent Typo

The mediaIndependentOutHighCapacityOctets object description
clause should refer to the 'number of octets in all packets'
rather than the 'number of packets'.

2.2) Network Utilization

The etherHistoryNetUtilization object in the RMON MIB (N),
reports the percentage of bandwidth usage. The behavior
of this object is not clear for 'complex' data sources such 
as switch backplanes or global VLANs within a switch (configured 
via the dataSourceCapsTable and SmonDataSource TC in RFC 2613), 

One vendor is setting the ifSpeed and 
etherHistoryNetUtilization objects to zero for these types of 
dataSources. Another vendor is setting the ifSpeed to the
highest ifSpeed in the aggregate (e.g., max(ifSpeed.X), where
X is each ifIndex in VLAN Y or connected to backplane Y).
This approach was not discussed in detail, but the WG seemed
to like it. However, this issue was not completely resolved
at the meeting, so the mailing list will be used to discuss 
this issue further. One issue: the basis for the utilization
formula (ifSpeed) will change as the members of the
VLAN or backplane change, 

3) Discussion of IF-TOPN MIB

The Interface TopN MIB (G) was presented by Dan Romascanu
and discussed by the WG.

Most of the discussion focused on the intent of the MIB itself.
The WG originally decided (at the last IETF) to develop a
MIB which was specific to interfaces, At this IETF, the
WG reversed this decision. This MIB is now intended as
a generic TopN feature for any numeric object, similar
to the generic history feature in RMON-2 (usrHistory group).

The current IF-TOPN MIB contains a mixture of interface-specific
and generic features.  Dan will write a new draft which 
focuses entirely on a user-configurable TopN mechanism.
The MIB will be renamed to the USR-TOPN-MIB module.
The current draft name will be terminated, and a new
draft called draft-ietf-rmonmib-usrtopn-00.txt will
be created.

4) Discussion of the DS-MON MIB (H)

The DS-MON MIB (G) was presented by Andy Bierman
and discussed by the WG.

The WG decided to remove the dsmonHostGroup from the 
MANDATORY-GROUPS section.

The WG discussed the idea of indexing the data tables 
differently. Instead of using the DSCP value itself,
an 'aggregation group' index can be used instead,
intended to model the per-hop-behaviors (PHB)
that are mapped to one or more codepoint values,
An NMS can configure these groups to contain an
arbitrary number of codepoints, and a codepoint must
belong to exactly one aggregation group.

Andy will write a new version of this MIB, which addresses
these issues.

5) Discussion of Virtual Data Source Issues

The WG discussed the notion of virtual data sources,
and the topic changed to support for Frame Relay data sources.

A virtual data source is some subset of the network traffic
present on a physical data source. The agent allows an NMS
to configure RMON collections on a virtual data source,
just like a data source for a physical interface. An
ifEntry must be maintained for each virtual data source,
for this purpose. The dataSourceCapsTable in RFC 2613
defines this procedure for Entity MIB and VLAN based 
data sources, via the SmonDataSource textual convention.
The SMIv2 does not seem to allow this TC to be augmented
with additional 'OID formats' for virtual data sources.

The WG would like to support DLCI based RMON monitoring in 
a standardized way. This issue was discussed in
some detail, and 3 options were presented to the WG:

 Option 1) just document current practices
           Tell vendors to elevate DLCIs to interfaces
           encode DLCI (Integer32) value into ifDescr, 
           ifName, or ifAlias object; ifType is 
           propVirtual(53)

 Option 2) extend SmonDataSource TC
           Add a new encoding for OIDs to the SmonDataSource
           TC that conveys the base interface index and
           the DLCI index (e.g., <foo.<ifIndex>.<DLCI>).

 Option 3) create a new Virtual Data Source MIB
           need to augment the SmonDataSource TC and
           dataSourceCapsTable to allow for DLCIs,
           other virtual circuits, etc.

The SMIv2 does not seem to allow for option 2.
Although the WG expressed some interest in option 3,
the WG cannot take on the new work required for option 3. 

Option 1 didn't seem complete enough.  The WG decided to ask some
members who also working on the Frame Relay Services MIB
to ask that WG to consider this issue. (Rob Steinberger has
already taken care of this action item.) 

6) Application Performance Measurement (I - L)

A summary of the Boston APM Interim meeting minutes was
presented by the WG Chair.  This was followed by
slide presentations of the APM Caps MIB (L). 
Steve Waldbusser then presented the APM MIB (K), 
and then Russell Dietz presented the TPM MIB (J).  
Refer to the slides for details on these presentations.

The WG then discussed various aspects of application
performance measurements. Active vs. passive testing,
terminology issues, and usage scenarios were discussed
in some detail.

After this discussion, the WG selected baseline drafts 
for this work item:

  - The APM Caps MIB (L) will be renamed to the
    PM Caps MIB so the name applies equally
    to APM and TPM capabilities. This MIB will
    be republished as draft-ietf-rmonmib-pmcaps-00.txt.

  - The APM MIB (K) will be republished as
    draft-ietf-rmonmib-apm-mib-00.txt

    The TPM MIB (J) will be republished as
    draft-ietf-rmonmib-tpm-mib-00.txt

The WG also discussed the need for protocol directory
extensions for application verbs and meta-protocols.
Some WG members have been tasked to produce a first
draft this document.  

The WG does not wish to document hundreds of application
protocols, since this approach did not scale well for
the Protocol Identifiers document (RFC 2074).
It is hoped that mechanisms can be developed which
allow an agent to describe the extensions it supports
in a self-referencing manner. This works best
for string based verb identifiers, but can also be
applied to the 'labels' of numeric or other types
of verb identifiers.

The WG must also  address the need for persistent
identifiers for usernames, DNS names, etc. 
The temporal measurements in the APM and TPM MIBs
will be much more useful if the endpoint identifiers
did not change meaning over (short-term) time.

The WG Editor has been asked to look into this
issue and propose some solution.  The WG should
discuss this topic on the mailing list.

7) Administrivia

The WG discussed 2 administrative topics at the end
of the meeting,

 - interim meeting
   The WG decided many months ago to have a 2 day 
   interim meeting on the west coast of the USA, in mid-May.
   This meeting has already been announced to
   the IETF and the RMONMIB WG. It will be held
   May 15 and 16 at the Sir Francis Drake Hotel
   in San Francisco, CA. See the rmonmib mailing list
   for details.

 - mailing list host
   The WG mailing will be moved to the IETF hosted
   servers in the near future, Current subscribers
   will be notified before the change, and will not
   have to re-subscribe to the new list.