WG Action: Next Generation Structure of Management Information WG (sming) to conclude

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

 



The SMIng WG has been closed. The WG could not get to consensus
on a new version of the SMI language and even today, some 2.5
years after being chartered, there seems to be no consensus
and/or wide community support to drive to conclusion on a
deliverable.


- In 1999 and 2000, the IRTF NMRG did a lot of work on a possible
  SMIng (Structure of Management Information Next Generation).
  The research group developed a complete new SMI language spec
  and also had implementations. In addition there were tools
  to convert to/from SMIv2 and tools to convert to/from other
  languages like IDL etc.


- In November 2000, the SMIng WG was chartered. The WG was 
  specifically limted to NOT try and develop a grandiose mother
  of all information/data modeling languages. Instead it was 
  chartered to enhance the SMIv2 and to try and cater for a
  data modeling language that would re-merge the SMI and SPPI
  back into one advanced data modeling language. The resulting
  Management Information Modules would be able to be used by
  multiuple protocols, specifically SNMP and COPS-PR.


- The WG was chartered to consider the NMRG SMIng work as base
  to start from. A requirements document was the first deliverable,
  to try and make sure it would address the most pressing needs
  but also to limit it to doable and implementable requirements.


- The WG was originally chartered to finish by March 2002.


- The WG got bogged down (initially) by defining the requirements.
  It took the WG 8 months (a year before the RFC showed up) to
  deliver the RFC (3216) that describes them (renamed to "objectives"
  because people worried that "requirements" would be used to "force"
  various "requirements" into a possible solution, even if the
  actual solution would turn out to be un-implementable or
  impractical. In hindsight, I think these were serious signs
  of trouble in the WG.).


- The WG did not show (even rough) consensus on using SMIng as the
  base to start from. So the WG then went into a "submission of
  proposals" and later "selection" process. I believe this did 
  bring good discussions into the WG, but it took a long time
  before a consensus was formed. The consensus I believe was to
  "merge the SMI-DS and the SMIng proposals".


- However, when trying to work from that consensus, the WG seemed
  to have tired of the work. Not enough volunteers could be found
  to actually write the "merged" document set and to drive it
  to conclusion. During that process, it turned out that various
  core contributors continued to have serious concerns about some
  of their original ideas/concerns not being considered/addressed.
  There was also a feeling (by some) that the compromise solution
  would not be coherent and that it would lack a clear design.


- It all culminated to an "un-volunteering" at the 55th IETF in
  November 2002. Instead of various members volunteering to help
  finish the documents that were agreed to by the WG, one of the
  volunteers "un-volunteered" some of the documents he had earlier
  volunteered to work on. The WG was warned soon afterwards that
  some better consensus, energy and effort towards a good result
  was needed in order to continue its work.


- We are now 2 years and 4 months after the WG started and we have
  seen no new initiatives in the WG for the last 4-5 months.
  What a shame that we have wasted more than two years without
  any results. The IETF NM community clearly has lost interest
  to try to get consensus and continue work based on such consensus.


- If the NMRG wants to publish the original SMIng docs as experimental
  or informational RFCs, they should go ahead and do so, so that we
  at least keep some record of a specification and implementaion
  that once was available.


Bert Wijnen
co-AD for the IETF Operations and Management Area


[Index of Archives]     [IETF]     [IETF Discussion]     [Linux Kernel]

  Powered by Linux