The Wage and Hour Division (WHD) at the Department of Labor (DOL) has been relying on an aging system, called Wage Hour Investigative Support and Reporting Database (WHISARD), to support its labor law enforcement activities. The entire system needs modernized, which includes replacing the existing desktop-only user interface (UI) with a new web-based UI. By careful inspection of the code and experimentation, we created a deep "spike" solution that allows a small fraction of the system to be implemented with a web-based system using in-situ business rules and stored procedures. This spike served as a basis for planning the entire rewrite of WHISARD, thus de-risking the whole project.
The existing technology behind WHISARD was ancient and nearly undocumented. In order to have a functioning spike, a complete and valid environment would have to be duplicated. Because the data was sensitive and potentially subject to litigation, we'd have to work entirely with artificially-created data.
After setting up a DB2 instance with artifical data and extensive study of the stored procedures, we were able to interface to the database with a simple Python program that implemented a web service interface. This allowed a simple but possibility-expanding, browser-based demo.
The spike solution dramatically reshaped the DOL's acquisition plan, including cost estimates, by proving that an API-encapsulation approach could work and be expanded at minimal risk.