Using Pattern-Based Architecture Reviews to Detect Quality Attribute Issues - An Exploratory Study

Architecture reviews are effective for identifying potential problems in architectures, particularly concerning the quality attributes of the system, but are generally expensive. We propose that architecture reviews based on the architecture patterns and their interactions with quality attributes can be done with small effort. We performed an exploratory study to investigate how much time and effort is required to perform such a review, and how many related issues it uncovers. We performed nine architecture reviews on small systems, and recorded the time and effort spent, and the number of issues identified. On average, a pattern-based review took less than two person-days of effort and less than a day of calendar time. The median number of issues identified was three, one of which was major. We recommend that where extensive architecture reviews are too expensive, a pattern-based review can be done with small effort and time.