Re: Time commitments for MIB review

[Harrie Hazewinkel <harrie@lisanza.net>]

Harrington, David wrote:
> Hi Bert,
>
> How about setting a minimum yearly commitment to remain a mib doctor?
> I'm not sure how many mibs are produced yearly or how many mib doctor
> reviews you want for each, but I see little problem with committing to
> perform, say, six reviews per year.
>
> I have concerns about a two-week deadline. My company pays my salary and
> sponsors my participation, and often work due for the company must take
> precedence over mib doctor reviews. I would have difficulty guaranteeing
> a two-week turnaround whenever you want it. This is especially true for
> mibs that had no mib technical advisor to ensure that a mib is not
> poorly designed and then a huge poorly designed mib is presented for mib
> doctor review. The mib doctor task for a mib that needs massive changes
> is much more difficult than a review for a well-designed mib. The
> Printer MIB jumps to mind, and recently the DHCP mib warranted a massive
> overhaul.

I beleive it would help if a MIB doctor is involved earlier.
That makes the 2 weeks period easier, since various things are
already addressed before and it become more a checklist issue.

> I think the comparison to RTG may not be valid. Does the RTG directorate
> perform review of large numbers of non-routing-related documents? While
> mibs are surely NM-related, the thing being managed by a mib is seldom
> an NM-related protocol. MIB Doctors need to have some understanding of
> the management target, not just mib-writing skills, and that takes time
> to develop. I think this is less of a requirement for the RTG reviews.

This is indeed also my concern. Doing a MIB review based on the 
so-called CLRs is quit easy. Just follow them, but doing a review
with respect to the technology as well is different.
Also if there is a 2 week dead-line a MIB doctor not understanding
that technology in details is difficult. A reason why I do not always
jump forward as a reviewer.

> I have concerns about being "assigned" mibs. It is much easier for me to
> justify the time to perform mib doctor review if it relates to my
> company's business. I think I can better judge that than you. 

Assignemnt is not directly bad, as long as they are assigned with
respect to the technology understanding of the MIB doctor. For instance,
I personally was never involved heavily in routing, so doing a review
there is difficult, but application level I have better understanding.

> MIB Doctors often do mib reviews even before you ask, and often the same
> mibs may be reviewed by multiple mib doctors, leading to some confusion
> about who is "responsible" for doing the mib doctor review. Should we
> have some type of volunteer registration mechanism to better coordinate
> this? I would find it helpful (and I'm sure editors would too) to know
> just who is "officially" responsible for reviewing which mibs.

I believe her we can distinguish two kinds of reviews. 1) A MIB doctor
reviews it since it is usefull for his work or 2) a MIB doctor reviews
it as an IESG review. Here I believe that 2) would be responsible
and 1) is just like anybody else out there.

> A related concern is that the IETF has swung decidely in a
> service-provider-focused direction over the past few years, and to a
> degree away from non-SP-related IP-centric data networking technologies.
> Are there patterns to the unclaimed mibs in your queue? Maybe you need
> to recruit and train some new mib doctors with a more SP-focused
> background.

Maybe some web page with which MIB modules need to be reviewed or
which WG need some help as well as which area is each MIB doctor
more familair.
Although, this is maybe not enough. I can think of the rserpool
situation. I volunteered to help, but to me it looks almost
like the WG stopped working on it. Not sure if they expect me
to write the complete MIB module or that it simple became not
important to the WG.

> I would also like to suggest that all WGs developing mibs should be
> required to have a mib advisor to perform design reviews, to ensure that
> the design is kept reasonable, and we should develop some BCP mib doctor
> guidelines for this as well. A mib advisor should probably be limited to
> maybe a one-year term, so they don't end up designing the mib themselves
> and then performing design review as well.

I beleive a MIB advisor can serve longer, but a final review must then 
still be done by a different MIB doctor. Just to make sure that things
do not slip in over time.

cheers,
Harrie