| reehouse
More detailed comments from Brian Johnson at Cutler-Hammer:"I highly recommend tRelational and DPS from Treehouse Software for data propagation from Adabas to Oracle (or any other RDBMS). We previously used a combination of PLEU and SAS/Access to do the job, but maintenance became a nightmare, DASD storage and processing was too costly, and the whole process was not as efficient as it could have been.
tRelational (and tRe-PC which runs on Windows but talks to tRelational on the mainframe via Treehouse Remote Access) makes it easy to map out the relationship between your Adabas files (which tRelational gathers metadata about from FDTs and DDMs) and relational tables, and maintenance to the model after-the-fact is intuitive and quick, even for people not trained on Adabas concepts. This is where you define your field mappings, external transformation routines (many are provided, but you can code your own), and information about your relational database. With a single batch job, you can generate all of your DPS parameters (control data for the bulk loader, FTTE data, etc.).
I really like the fact that DPS doesn't use Adabas' resources, but rather reads the ADASAV directly for materialization and the PLOG directly for propagation (no more PLOG extracts using PLEU or ADASEL!). This really means that it can accomplish the job of providing you CTL or MAT files ready for bulk loading or SQL for updates/inserts/deletes in much fewer CPU cycles than you can do yourself, plus all the problems you'll have to deal with one way or the other when interpreting PLOG data (i.e., backed-out transactions, transactions that span PLOG tapes, transaction referential integrity) are already handled intelligently by DPS.
I installed Oracle Client for MVS to accomplish the task of adding a final jobstep that takes the resultant SQL dataset and applies it directly to an Oracle 8.05 database residing on HP-UX Unix via SQL*Net. We looked at many different products recently to determine whether we should stick with SAS/Access or use something else. No other product anywhere can handle changes in Adabas, which means all other solutions require the use of PLEU or ADASEL. That is a severe handicap to your process because not only is it inefficient use of CPU and DASD to do so, and not only do you then have to deal with the problems of PLOG processing I mentioned before, but PLEU and ADASEL don't really give you all the data you need if referential integrity is required (namely the issuance of BT/ET). Needless to say, I believe tRelational and DPS is the best solution available for your data propagation needs." Brian Johnson
|
Download Brian's PowerPoints from his presentation at the Midwest Region SAGAGROUP User Group meetingChronicling four attempts over a period of six years, Brian shows how Cutler-Hammer is heading in the right direction with their Data Warehousing effort using tRelational and DPS. During their first three endeavors, they encountered many obstacles, including difficult maintenance requirements and inefficient PLOG extraction and transformation, as well as PLOG, ad hoc update, and bad data problems. By using tRelational and DPS, these difficulties and many others have been eliminated, and Cutler-Hammer is now able to support their Data Warehousing projects. Download Brian's PowerPoint presentation. |