An Expert System for Correcting JEE Performance Anti-Patterns

CRASSO, M.; ZUNINO, A.; CAMPO, M.; MORENO, L.

Mar del Plata

Simposio; 10th Argentine Symposium on Computing Technology (AST2009) - 38th JAIIO; 2009

SADIO

Java Enterprise Edition (JEE) middlewares have been successfully employed to build enterprise software systems, while achieving security, flexibility and reliability. However, achieving performance requirements through these middlewares has been identified as challenging. We propose a novel approach for automatically detecting performance problems in JEE-based applications and suggesting how to correct them. The idea is to automatically detect well known performance anti-patterns and suggest how to adapt concrete solutions to the performance problems of an application, from anti-pattern descriptions. Experimental evaluations show that after correcting the anti-patterns found in three JEE reference applications, their response times represent a 40%, 76% and 61% of the original response times.