The technical challenges with any information integration all stem from the fact that the application accesses heterogeneous data (VSAM, IMS, IDMS, ADABAS, DB2, MSSQL, and so on) that can even be in a non-relational hierarchical format. Some of the technical problems include:
The flexible file definition feature used in COBOL applications in the existing system will have data files with multi-record formats and multi-record types in the same dataset — neither of which exist in RDBMS. Looping data structure and substructure or relative offset record organization such as a linked list, which are difficult to map into a relational table.
Data and referential integrity is managed by the Oracle database engine. However, legacy applications already have this integrity built in. One question is whether to use Oracle to handle this integrity and remove the logic from the application.
Finally, creating an Oracle schema to maximize performance, which includes mapping non-oracle keys to Oracle primary and secondary keys; especially when legacy data is organized in order of key value which can affect the performance on an Oracle RDBMS. There are also differences in how some engines process transactions, rollbacks, and record locking.