This paper presents a failure analysis of the Java Virtual Machine providing useful insights into the nature of reported failures and to improve the understanding of its dependability aspects. Failure data is extracted from publicly available bug databases, where developers and users of Java applications usually submit failures/bugs.

Presented results clearly indicate that much more efforts have still to be done in order to improve the dependability of the JVM. In particular, the conducted analysis revealed that i) builtin error detection mechanism are characterized by a low coverage; ii) the JVM does not achieve the same levels of dependability across different platforms iii) developers have to pursue a tradeoff between performance and reliability. Finally, code fragments reproducing failures submitted in bug database are injected into Java Applications. Preliminary results show that often these faults could be removed changing the environment of the JVM.