Jump to: navigation, search

Cosmos Architecture Meeting 24-July-08

Revision as of 10:07, 24 July 2008 by Dlwhiteman.us.ibm.com (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

These are the minutes for the Architecture Meeting for 24 July 2008.

Attendees

  • David
  • Paul
  • Jack
  • Jason
  • Mark McCraw
  • Srinivas
  • Ray Ellis
  • Jimmy (definitely not lurking)
  • Martin
  • JT (lurking)
  • Bill
  • Hubert

Agenda

  • Status update on i12 candidate build
  • Ray Ellis of IBM: An alternate implementation of the CMDBf framework using XSLT (20 minutes) (details)

Meeting Minutes

i12 Candidate Driver Status

  • JT has a bug fix checked in that will require a respin, and will contact Jagmit for this
  • Srinivas mentioned 3 problems, all of which are likely due to the bug JT mentioned
  • Martin mentioned a minor bug that is not a stop ship issue and will create a bug report for this
  • David reported that the RM and ME issues found early yesterday have been resolved in the latest build

XSLT in CMDBf framework

  • Ray Ellis of IBM presented an approach he's prototyped for using XSLT to process larger grained registration and query requests. This has the benefit of using the XSLT standard, and allows for a more transactional approach, leveraging the scalability support of relational databases, and eliminating the post processing done in main memory by the current implementation. The templates would allow for plug points for custom code.
  • Hubert mentioned this is complementary to the pluggable approach he had previously presented
  • Jimmy indicated a lot of interest in this approach, and CA would likely adopt it in their own MDRs if it were present.
  • Bill requested an example of a query request template. Ray says he hasn't completely worked out the query scenario, but that's mostly just been a time issue, as he has a previous attempt based on Antlr that he can translate to the XSLT approach. Ray will post that example in a few days.
  • The group felt that, despite the fact there is not a lot of remaining time before COSMOS 1.0, we should give strong consideration to taking on this effort for the release, since it would remove a lot of barriers to adapting real-world data stores to CMDBf. Jimmy volunteered to be part of the discussion in working out the details.