Re: WG split: why and how?
Alex Zinin <[email protected]>
| Newsgroups | gmane.ietf.ppvpn |
|---|---|
| Message-ID | <[email protected]> |
Folks-
As promised (though 11:40-ish my time), I tried to quickly bring
together information that would clarify some confusion I noticed
during the discussion about the work split. I also tried to come up
with answers to questions that I thought might be interesting.
Hope this is helpful.
Regards.
Alex
1. Procedural background:
1.1 Relevant quotes from RFC 2418:
Section 2:
a) "...A working group may be established
at the initiative of an Area Director or it may be initiated by an
individual or group of individuals."
b) "Working groups are typically created to address a specific problem or
to produce one or more specific deliverables (a guideline, standards
specification, etc.). Working groups are generally expected to be
short-lived in nature."
c) "Upon completion of its goals and achievement
of its objectives, the working group is terminated. A working group
may also be terminated for other reasons (see section 4)."
Section 4:
"If, at some point, it becomes evident that a working group is unable
to complete the work outlined in the charter, or if the assumptions
which that work was based have been modified in discussion or by
experience, the Area Director, in consultation with the working group
can either:
1. Recharter to refocus its tasks,
2. Choose new Chair(s), or
3. Disband.
If the working group disagrees with the Area Director's choice, it
may appeal to the IESG (see section 3.4)."
d) "The formation of a working group requires a charter which is
primarily negotiated between a prospective working group Chair and
the relevant Area Director(s), although final approval is made by the
IESG with advice from the Internet Architecture Board (IAB). A
charter is a contract between a working group and the IETF to perform
a set of tasks."
1.2. WG split is not covered in 2418
From the process perspective, splitting of PPVPN work is
considered as 1) termination of the existing WG, 2) formation of
the new WGs, and 3) relocation of the WG documents to the new WGs.
2. Data about PPVPN WG:
BOF'ed : Pittsburgh, Aug 03, 2000 (NBVPN BOF)
1st meeting : San Diego , Dec 14, 2000 (PPVPN BOF/WG)
WG drafts : 17 (as of May 09, 2003)
Submitted to IESG : 3 (L3 framework, L3 reqs, generic reqs)
Approved, in RFC-ed queue : None (as of May 09, 2003)
RFCs published : None (as of May 09, 2003)
3. Problem statement.
As seen by ADs: slow progress on L3 and L2 VPN work (see section 2
above) caused by comparatively wide problem space with unusually
big number of correlated technologies (L2, bridging, tunneling,
routing, signalling, etc.) that in turn results in a big number of
tasks and documents within the WG. Overloading of the WG resources
(management cycles, mailing list, and face-to-face meeting time)
causes slow progress on the existing WG documents that (because of
the number of incomplete items) keeps continuing VPN technology
developments outside the WG and makes the process within the WG and
its results less relevant to the community and the industry. Unless
addressed, the situation is expected to get only worse.
4. Summary of the solution under consideration
1. Put L3 and L2 work in separate WGs
Analysis of the L3 and L2 remaining and potential future work items
has shown that there is a sufficient amount of work in each area to
warrant a separate WG. Smaller and more focused WGs tend to be
more efficient in terms of achieving consensus and completing
their tasks.
2. Focus the charters to ensure progress on items already in
the pipe; expand the charters and add more work items as
things progress
5. Goal of the discussion on the list
Collect community feedback on the solution and the charters of
the proposed new WGs.
6. FAQ
6.1 How is relocating the work to the Internet area related to
the described problem and the proposed split?
It is not. Relocation of the VPN work to the Internet area
is an independent action. However, since the question of
reorganizing the work is being considered, it is logical to
do the move at the same time.
6.2 Does the IETF process require consensus within the WG on the problem
statement and/or the proposed solution?
Formally, the ADs or the IESG do not need to obtain WG consensus
on forming, disbanding, splitting a WG, or any other WG management
actions. However, it is normal for the ADs and the IESG to consult
the community before making the decisions. This is what happened
before the discussion was started and essentially what is
happening on the list now.
6.3 Have you made the decision already or are you still open to
feedback from the community?
We came to the WG with a strong opinion that a split is needed,
however, the decision has not been made yet and we are open to all
opinions.
6.4 If the decision has not been made yet, why did you start the
discussion on the charters of proposed new WGs?
This was done to make sure the community has enough information
to understand what the results of the split would be, and to
collect feedback on the charters themselves.
6.5 What would happen to the documents that span both L2 and L3
VPN areas?
They would be put in one of the new WGs.
6.6 How would you ensure coordination between the two VPN groups?
Coordination between different WGs is a well-known task and is
usually accomplished through direct communication of the WG
chairs, ministration by ADs, and by formation of technology-
specific directorates. Formation of the VPN directorate is being
considered.
6.7 Have you considered other alternatives to the split?
Yes. The alternative to the split would be to organize work within
the WG in the form of design teams and have more than one WG
session at the face-to-face meetings. However, the WG already has
a number of DTs inside of it, so we can consider that this method
has been tried already. Increasing the number of WG sessions alone
is hardly going to help as majority of work is done outside the
face-to-face meetings, in fact, consistent need for more than one
session maybe considered as an indication that the WG has more
than enough on its charter.