Summary of conference call July 20, 2004 In attendance: Larry Flon Hemant S Kolwalkar Brian Collins Susan Malaika Jan-Eike Michels Norman Paton Tom Roney Reference for this meeting: From: Larry Flon To: cgs-wg@gridforum.org Cc: Todd Guay , rreich@veritas.com Subject: [cgs-wg] Telecon Tuesday 20-July and Documents to Review Date: Fri, 16 Jul 2004 11:19:57 -0700 I have uploaded to the gridforge web site a ZIP file contributed by Susan Malaika of IBM. This ZIP file contains a UML model, created in Rational Rose and published as HTML files, of a relational database schema. under Documents/Meeting Materials and Minutes/SRIM Drafts. The file is Contrib_UML_SQL_MODEL At the Tuesday telecon, we will devote most of the time to reviewing this model. So all of you who plan to join the telecon, please make sure you can access, and download the ZIP file. Susan and her colleagues will walk us through the model. --------------------------------------------------------- Hemant traversed the Main SQL Model, which consisted of class diagrams for Schema, DataTypes, and Tables. He also reviewed the class diagram for Inheritence. Every class derives from the Main SQL Object. This was a very high-level view of the model. One issue is going to be the creation of MOF. Can we produce it? Is this the model we want to go forward with? Do we use the model exactly as it is? It will be a mostly manual process to create the MOF, and we don't want to do that until we are comfortable with the model. This goes back to use cases. Can we perform function X? Will have to take a deeper look. The DAIS group will do this, and annotate their use cases for discussion during our next teleconference in two weeks. The DAIS group is planning another "specs" telecon. What was the derivation of this model? Just started from scratch and based upon the SQL99 model. The CWM people did the same thing. Should we perhaps look at the two models? Susan thinks this one a better match for relational data. It matches very closely the SQL standards. IBM's SQL-Standards expert, Jan-Eike Michels, verifies that the representation of this UML matches closely to SQL. May want to discuss later: special vendor extentions. Will want to construct a list of future extenstions. The diagram called inheritance, that lives below tables . . . Does that contain all of the classes? Yes, but it does not contain the associations and attributes. How many classes are there? Something close to 40. So, the next important step is to look at the use cases and make sure that the functions that we want are accomplishable with this model. For discovery, in particular, is this going to work? Plan for next teleconference: The DAIS team will look through the model to do an internal review, and discuss the use cases and send a note with the scenarios annotated. When we have confidence that this is going to meet the requirement, we will run it by other groups for review. There was some discussion about a new DATA-DESIGN team in GGF. Not as detailed as DAIS. Looking at DATA as a whole.