The challenge.
A major insurance company based in Switzerland was adopting an
Oracle-based data warehousing and data mart solution.  They lacked
the expertise and tools to redevelop existing COBOL / DB2 code and
ETL logic to enable the seamless transition to an Oracle environment. 

Our solution.
The generated COBOL/ DB2 code from the PRISM tool had been
modified during the lifecycle of the legacy data warehouse. So we
decided to use the generated COBOL / DB2 code for understanding
the ETL process rather than the PRISM specifications.

NIIT Technologies used in-house developed code mining tools to parse
the generated COBOL code and derive out the ETL transformations and mapping specifications. The ETL logic was redeveloped in an Informatica/ Oracle environment.

The DB2 data warehouses and data marts were migrated over to the
new Oracle-based database environment Regression testing was adopted as the test strategy. The original COBOL based ETL processing streams were run in the legacy AIX environment end to end. The new Informatica based processing streams were run for the same host extracts that had been fed into the legacy side. Scripts were developed to compare the end point updating taking place in the legacy DB2 and new Oracle databases. Acceptance testing was defined to be successful is the end point data updating happening in the Oracle databases was the same as that occurring in the legacy DB2 data warehouses and data marts.

The results.
The client’s processing requirements were efficiently redesigned to
the new Oracle architecture. Host systems were relieved of high-impact spikes in demand which increased overall systems performance. A full regression testing strategy was adopted which continues to help the client identify and eliminate data inconsistency and other disabling issues.