A method for the manual extraction of business rules from legacy source code

One of the main barriers to the successful migration of legacy systems is lack of information about the system to be migrated.With the complexity of modern information systems, it is rarely the case that an organisation has a complete end-to-endunderstanding of its systems, business processes and information. This is particularly true of the business rules enforced by suchsystems, which are often poorly documented (if at all) and incompletely understood by those who own, use and maintain suchsystems. These business rules must be migrated, along with the other system functionality, but this is difficult when they areburied deep within the source code of the system. In this paper, we propose a method for the discovery of business rules throughinspection of the source code of legacy systems. We also describe the results of a comprehensive evaluation exercise undertakenwithin BT, in which the method was applied to a legacy system about to undergo migration. The results of this study indicatedconsiderable success for the method in extracting valid business rules, but also highlighted some weaknesses. Most significantly,the cost of applying the method is a factor of the size and complexity of the system being analysed, and not of the number ofbusiness rules elicited. In order to address this, we propose a cut-down version of the method that can be used to produce a quickestimate of the likely yield of business rules from a program. Together, these two forms of the method aid personnel engaged insystem migration projects in making maximum use of the scarce resources are available for business rule discovery.