RFC on object/data modeling & CASE tools

From: Michael Wirth <mcw_at_cs.rice.edu>
Date: 27 Jan 1992 19:57:00 GMT
Message-ID: <1992Jan27.195700.4150_at_rice.edu>


REQUEST FOR COMMENTS ON OBJECT/DATA MODELING & CASE TOOLS I am interested in your comments on the following subjects:

  1. Appropriate, practical(!) CRITERIA which can be used to measure and select a data modeling and CASE tool set from current, commercially available alternatives.
  2. Your EXPERIENCE with such tools, particularly in heterogeneous "real world" environments like the one described below.
  3. Your ADVICE on how such tools can be used for near-term practical problems (such as mapping data elements from an old hierarchical database to various relational schemas), while still serving long-term needs in building a highlevel conceptual (i.e., object) model.
  4. Your comments on the appropriate METHODOLOGY which can span both our nearterm, tactical data transformation needs and our long-term strategic data modeling needs.
  5. Your comments on what TRAINING is appropriate for our team members.

BACKGROUND
My company has several large (5GB+, several million records) complex data sets which it currently maintains in a hierarchical, mainframe DBMS (CCA's Model 204). The NEAR-TERM problem, on which several of our groups is working, is to complete data dictionaries for several of these data sets, reverse engineer them to extract higher level structure, integrate the data sets to the extent possible, map the current data to various relational schemas (dictated by our clients) and deliver the data in those relational schemas.

The groups which are doing this work have extensive experience with the data, M204 and other mainframe DBs, and a little RDBMS experience, but no formal data modeling or CASE experience.

Another group (mine) is much farther out on a limb, technically speaking. We've developed a class library in CLOS (in the style of MacApp and Lispview) which allows us to assemble applications which are fully object oriented, but use relational DBs as backing stores for the objects. Given an object model for a domain and a "reasonable" relational schema for its data, we can map the object model to the relational tables to provide point-and-click access to the data. We're also experimenting with object DBs.

My group is, therefore, very interested in having the work of our tactical groups couple with my group's strategic goals (e.g., wouldn't it be nice if a tool existed which allowed us to map low-level data elements from a hierarchical DB to various relational table layouts, while capturing the high-level design in an object model, in the manner of Rumbaugh, et al's Object Modeling Technique).

As usual, I'll be happy to summarize responses (to which newsgroups?).

Mike Wirth
713-850-9295 (phone)
713-963-0085 (fax)
mcw_at_cs.rice.edu

PS: I have deliberately not named our company or industry in this posting. I'll be happy to provide that info to individual respondents. Received on Mon Jan 27 1992 - 20:57:00 CET

Original text of this message