Jeff Parker, MIB update, draft-parker-isis-wg-mib-xx.txt
--------------------------------------------------------
Presented updated MIB draft. Several new values has been requested
and new variables will be introduced, especially the possibility
to override hello padding and large circuit id to support more than
255 circuits. Discussions happened on the mailing list and conclusion
drawn that defaults should be removed from the isisSysTable since
their semantics were very confusing. This remains to be done in the
next version of the draft.
Tony Przygienda, IPv4 update, draft-ietf-isis-wg-over-ip-xx.txt
---------------------------------------------------------------
Mixed Ethernet IP and OSI encapsulation has been dropped from the draft.
The artificial Ethernet encapsulation has been dropped as well after
careful thinking. Dave Katz made the comment that we can use the local
bit to make it work. Requires some more thinking and possibly another
version. Fragmentation issues have been carefully looked at based on
several comments and no further input has been given from the group.
Introduces wider metrics, L2 to L1 leaking to solve MED problems and
traffic engineering parameters for links. Proposed a TLV format. The
intention of the draft is for the new TLV to replace 128 and 130
down the road. Question has been asked why the name of the node
is not included into this TLV as well.
Several comments came from the group:
x bandwidth is expressed in %, why don't we carry
absolute numbers ? This seems to be under discussion.
x currently available bandwidth is carried instead of reserved
because of rounding problems.
x Maximum link bandwidth could be a
complete vector as well, not only one number. Preemption will
have problems with the encoding due to the assumption that only
level 0 is non-preemptible. Parallel changes should be done in
ospf draft.
x With the up/down bit, do we need other proposals for L1/L2 leaking ?
The answer is yes, deployment issues and not clear whether everyone
will use the presented draft.
x IPv6 issues have not been addressed.
Tony Przygienda, draft-ietf-isis-l1l2-00.txt
Tony Li, draft-ietf-isis-domain-wide-00.txt
--------------------------------------------
Both drafts proposed to solve the same basic problem of lack of globally
optimal routes and meaningfull IGP metrics for MEDs with different
encodings. After the presentations, technical discussion dissolved
into comments on possible loops forming without concrete examples
and more discussion on private and general mailing lists has
been encouraged. More presentations and progress is expected.
Dave Katz, 3-way hello, draft-ietf-isis-3way-00.txt
---------------------------------------------------
Dave presented the draft. One question was asked of whether it is clear
in which states the optional part needs to be included and the answer was
that the state machine covers it.
New Milestones Presentation and Discussion
------------------------------------------
new milestones have been presented and will be approved and posted
x recommendations to ISIS in ISO are not a deliverable anymore
x comments on v2 isis found a person that volunteered to compile it
x isis traffic engineering is submitted
x hmac md5 is submitted, will undergo revision. Ran Atkinson spoke
up and his concerns have been acknowledged.
x isis over ipv4 submitted, will probably have more revisions
x more than 255 adjacencies submitted, should go forward to informational
x optional checksums submitted. Waiting for more implementations
x MIB submitted, still in revisions
x improvements of core ISIS protocol are not a milestone anymore
x submit traffic engineering as informational
x submit isis over ipv4 as informational
x hmac md5 advancing not possible to submit at this point in time.
x optional checksums informational moved to Nov '99
x review priorities and milestones in Mar '2000
x move ISIS MIB as proposed standard
x traffic engineering MIB. Unclear whether should be done in ISIS.
If a new working group is formed, probably rather there.
Question was asked as of when is the earliest point at which the workgroup
could be shut down. This should be brought up at the next discussion about
scope and milestones.
At the end of the session, additional traffic engineering discussion items have been
presented by Ajay Kachrani without a draft and led to discussion.
Current fault-protected bandwidth reserved, current link utilization and
probability of packet drops have been proposed. Major comment was that this
has been considered early and dropped since no use was seen in terms of
computation/algorithms. As well, link utilization cannot be used safely
to drive computations without major oscillation issues. Ultimately, the feeling
was that without a proposed algorithm, the parameters are not necessary and
don't need standardization. Recommendation has been given to submit a draft with
a justification for the parameters beyond the desire to use the information for
management and data collection purposes.
