In my previous blog, we talked about the benefit of making subsets of test data from live production applications and masking them to address cost and security issues.  When the applications have simple data models where subsets can be made using a simple query on a few tables, the need to implement or purchase a solution may not be warranted.  When dealing with complex custom or packaged applications such as Oracle E-Business Suite or SAP, functional test cases are typically organized by business processes, organization, time, or a combination of each.

Complex custom applications or packaged applications contain data for multiple business processes, such as Accounts Payables or Receivables, Sales & Distribution, or Payroll.  Developing a SQL query that selects a complete subset for each of these processes for a particular business unit or geography and then masks the data while insuring the test application will continue to function is NOT a trivial task.  It requires a detailed knowledge of the data model – including all database constraints, primary and foreign key relationships, data dependencies on programs running in the application tier and any inter-application dependencies via database links or other interfaces.  It is possible to develop internally, but the time and effort required to develop and test makes the cost benefit of subsetting moot.

Unfortunately, Oracle and SAP provide limited subsetting or masking functionality or support for their customers.  These ISVs rely on partners such as Informatica, to provide preconfigured solutions for their applications that make sure masked data subsets can be created in such a way that the applications continue to function with no loss of data relational integrity.  Informatica’s Data Subset and Data Privacy for Oracle, PeopleSoft, Siebel and SAP Applications provide testing teams the ability to configure subsetting and masking policies through simple point and click.  Adding custom tables or altering standard subset criteria can be done through a GUI.

Creating masked subsets of live test data for packaged or complex custom database applications is one of those cases where the time and money saved by using prebuilt templates makes buying a product over developing something in house a no brainer.