Banking Case Study

Abstract

The online banking division of one of the world’s largest international bank’s launched a new version of its client website that served as the primary interface for its most highly-valued private banking customers.

This new, business-critical web site was experiencing performance degradation issues right from the start. The root cause of the problem could not be identified and, in order to avoid harming the bank’s reputation or possibly losing customers, the application servers were bounced twice a week on average.

After three months of repeated, but unsuccessful, attempts to find the root cause of the problem, the team turned to BTM. Using an innovative Business Transaction Management solution, the problem was diagnosed in a matter of hours and resolved it in a matter of days.

Why was BTM successful while other alternatives were not?

BTM revealed a gradual increase in CPU consumption for a specific group of business transactions. The root cause was diagnosed as repeated access to shared memory by the code generating the web's disclaimer page. This repeated access led to synchronization issues that, over time, became a bottleneck for the entire system. This kind of problem is completely undetectable by other tools.

BTM findings in this scenario included:

• 100% accurate proof-point that the problem occurred at the portal container and not at the backend application tier or one of the databases
• Proof that the issue was not related to container memory leakage, garbage collection or I/O problems
• The observation that the CPU consumption issue was only happening at certain transactions and not across the board
• A clear indication that the problem was not associated with certain users or user profiles

See diagram #1 below.

Diagram #1 - BTM monitoring showed an increase in CPU consumption for specific business transactions. This pattern was undetectable by any other tools and was key to understanding the true nature of the problem.