An empirical study on configuration errors in commercial and open source systems
Configuration errors (i.e., misconfigurations) are among the dominant causes of system failures. Their importance has inspired many research efforts on detecting, diagnosing, and fixing misconfigurations; such research would benefit greatly from a real-world characteristic study on misconfigurations. Unfortunately, few such studies have been conducted in the past, primarily because historical misconfigurations usually have not been recorded rigorously in databases.