Applying Test-First and Parallel Processing Techniques to ERP Data Conversion

Enterprise Resource Planning (ERP) systems provide a unified solution for streamlining the operations and flow of information across multiple divisions of an organization. Most of the functionality in ERP systems is master-data driven, and hence the migration of data from legacy systems to ERP is a very important activity during development. Such data must be correctly migrated so that integration and acceptance testing of specific functionalities can be performed. However, the high complexity of ERP systems, coupled with the large volume of data involved, make ERP data conversion an immense challenge for software practitioners. In this paper we tackle the ERP data conversion problem using a test-first approach. The approach seeks to identify and correct data mapping errors and anomalies prior to initiating the data conversion process. Fixing such errors early can avoid having failures occur during data conversion, thereby resulting in cost and time savings. In an effort to further reduce the length of time spent converting large volumes of data, we perform parallel execution of the data conversion program. We have conducted an industrial case study that applies our approach to two ERP projects with favorable results.