[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: SNMP improvements

At 08:37 PM 9/17/2003, Wes Hardaker wrote:
>>>>>> On Wed, 17 Sep 2003 11:10:22 +0200, "Wijnen, Bert (Bert)" <bwijnen@lucent.com> said:
>Bert> One other IMPORTANT factor was:
>Bert> Any work that had actually been done, documented and discussed
>Bert> did NOT reach any WG consensus as to where to go.
>I believe there was consensus at the end, at least in the EOS case
>(but it was also likely that not all the issues had surfaced yet with
>the latest work).  The SMIng case had consensus on most points as well
>(with a few exceptions).
>Bert> - volunteers to do the work (editing, leading the WG, reviews, 
>Bert> possible implementation and experimentation work).
>There was volunteers to do some of the work, just not all of it.  For
>EOS, you had the editors, WG chairs and implementors but less people
>were interested in reviewing it (aside from the implementors and
>editors).  Heck, 2 drafts had been published between the 2 IETFs just
>before it was shutdown at the second.  For SMIng there was
>significantly less support, unfortunately, as the core authors dropped
>the drafts.  Certainly the spirit in both groups was understandably
>>> The political environment in the IETF NM area has been pretty
>>> discouraging to people interested in SNMP. It has largely come down to
>>> if you are interested in SNMP your opinions are not welcome. 
>>> That is one reason why you see so many SNMP people in the netconf WG.
>Bert> As AD for the IETF NM and SNMP related work, I must take issue
>Bert> with the above statement. It is NOT the "political environment
>Bert> in the IETF NM area".  It is rather that the players seems
>Bert> unable to put enough energy into it and lack the willingness to
>Bert> come to consensus on the technical solutions.
>We're now in a position where my roadmap presented in SF isn't
>happening (at we're at the case I depicted with a dead log lying on a
>car).  Some area directors have advised some working groups that they
>should consider carefully how they want to do network configuration
>and whether they should use SNMP, COPS, or any other existing common
>technology to do so.  This has lead to the belief that WGs should do
>their own thing till netconf gets to a deployable state.  In the mean
>time, that leaves us in a state of either doing many different
>WG-specific solutions rather than using generic reusable mechanisms,
>or work is tabled until such a time that a generic method (netconf)
>fully spins up.  Both of these are evident in multiple working groups.
>In either direction, its a bad solution and we're now faced with
>fragmentation and stagnation.  Netconf is making great progress, but
>there is a long way to go before it's deployable and usable by other
>WGs.  It'll be 2 years minimum probably to get a semblance of a common
>data model together and to resolve the transport issues (let alone the
>protocol issues).  Fortunately, we will make it there eventually but
>there is a big change-over pain we'll have to go through first.
>If I sound pessimistic, it's because I am.  We've burned our bridges
>before we've even crossed the ravine, and the bridge forward isn't
>fully built.  I'm confident that we'll be ok in the distant future, as
>the netconf work is on a good track.  It's the years until then that
>I'm concerned about.

I don't share your pessimism.  First of all, the significant
changes to SNMP and SMI that didn't happen in the EOS and SMING
WGs would have taken years to finish and deploy.  Also, I
don't understand why "fixing" SNMP is so important anyway.
The industry has decided SNMP is good for monitoring and
notifications, and it does a great job in both cases.  I don't
see any compelling need to make significant changes to
the SMI or SNMP for these tasks.  I think SNMP applications
will coexist with NETCONF applications for many years to come.

>"In the bathtub of history the truth is harder to hold than the soap,
> and much more difficult to find."  -- Terry Pratchett