Using Defect Patterns to Uncover Opportunities for Improvement

Although no one is happy to find defects in their software, defects are introduced and removed continually during software engineering processes, and it is practically necessary to acknowledge, record, and analyze those defects to make progress toward higher standards of quality. The Orthogonal Defect Classification (ODC) methodology, first introduced by Ram Chillarege and his colleagues at IBM [Chil92, Bhan94], provides one approach to analyzing defects. The ODC methodology has been adapted and extended for use within Bellcore, and a web based tool set, called the Efficient Defect Analyzer (EDA), has been developed to support ODC use on a large scale. This paper outlines how EDA has been successfully applied, potential pitfalls, and some possible future directions.