Performance and scalability of EJB applications
OOPSLA '02 Proceedings of the 17th ACM SIGPLAN conference on Object-oriented programming, systems, languages, and applications
Cork: dynamic memory leak detection for garbage-collected languages
Proceedings of the 34th annual ACM SIGPLAN-SIGACT symposium on Principles of programming languages
Precise memory leak detection for java software using container profiling
Proceedings of the 30th international conference on Software engineering
Patterns of memory inefficiency
Proceedings of the 25th European conference on Object-oriented programming
Fast extraction of adaptive change point based patterns for problem resolution in enterprise systems
DSOM'06 Proceedings of the 17th IFIP/IEEE international conference on Distributed Systems: operations and management
Hi-index | 0.00 |
SOA environments are typically characterized by large number of frameworks. These frameworks stack over each other in the runtime infrastructure and result in deep stack depths and large number of objects being created, most of which, are short lived. Consequently, problem determination and performance tuning of such runtime environments is known to be an extremely difficult task, which requires experience and expertise. In this paper, we share our experiences working with such production SOA runtime environments. Through our experiences we try to find the answer to the following question: can problem determination and performance analysis itself be offered as a service in SOA environments? We note that, in practice Java language and the associated J2EE stack remains one of the most popular runtime environment for implementing SOA. Since Java provides structured runtime logs, we seek to find patterns in those logs that can be used to automate performance analysis and problem determination in SOA environments. We describe three performance problem case studies, each of which present unique performance problems in open source benchmark and production SOA applications. All the case studies highlight the complexity associated with automated performance analysis. However, we make the case that at least part of the performance analysis process can be automated and offered as a service.