Problem Determination in Enterprise J2EE Systems: Motivating Precision in Root Cause Analyses

A multi-tiered enterprise J2EE system typically comprises a number of distributed infrastructure components, 
including a web tier, application server tier, database tier and other components such as a corporate LDAP, 
Proxy Server, Policy Server and so on. Problem determination and root cause analyses in large distributed 
enterprise systems is often difficult and time consuming, requiring subject matter knowledge (and in many 
cases carefully planned instrumentation) to pin-point the root cause of failures prior to planning the required 
correction procedures to a given problem. Research in the area of problem determination and root cause analyses 
are of key interest to IBM and IBM’s competitors, on the basis that critical situations can be expensive to 
resolve and correct. Likewise, research in the area of autonomic self-healing techniques are equally pertinent, 
as solutions in this area improve high-availability and up-time percentiles in customer installations.; In this 
paper, we shall present a new methodology to allow systems’ engineers determine the root cause of problems in 
large distributed J2EE infrastructures. We shall describe an approach where failures in an infrastructure can 
be automatically correlated to the specific user actions that motivated the failure in real time.