Case study: Load Testing of an Oracle Siebel Application

Our client is a universal bank headquartered in a European country. It offers a broad range of products and services through its extensively developed retail, corporate, SME and micro banking business lines. Due to NDA restrictions we cannot explicitly mention the client name.

Background

While merging with another bank, the client switched to a different CRM system based on a newer version of Oracle Siebel Open UI. After the initial setup the system required load testing to confirm its performance parameters.

Project goals

The client had its own QA team experienced with performance testing. However due to the tight schedule, complexity of the test design and limited budget they sought for an external consulting service to establish initial setup for their tests. The goal was to choose a load testing tool, analyze the typical system usage scenarios and implement the selected user sessions.

Technical features

  • CRM system accessed by 3 different types of end users;
  • Oracle Siebel Open UI implementation;
  • Each user sessions consisted of 5 to 10 form fillings with each transaction requiring handling of a number of session-specific parameters;
  • For security reasons the site was available only through a special VPN and remote access connection to the client system within their LAN.

Solution

Load testing of any Siebel application is a challenging task, because the implementation of the client-server communication scheme is very complex and requires handling session-specific values at each step. There is no fully automated solution and the one merely facilitating the process is provided by LoadRunner with additional components supplied by Oracle. The license cost of this solution was 5 times higher than the project budget. That is why the client decided to use WAPT Pro and additional test design services from SoftLogica.

We spent 4 days for the initial analysis. As a result, we negotiated with the client 7 test profiles selected for implementation. One more week was required to complete and verify each of them and deliver to the client with full description of the parameterization added to the user sessions.

During the next week we worked closely with the client team providing help on the interpretation of the test results conducted with help of the created test sources.

The testing revealed several restrictions in the system configuration that prevented it from working as expected. One of them was the limit on the maximum number of active sessions - the application could not serve more than 30 simultaneous users. The reason was in the default server setting which limited the maximum number of connections for a single IP address.

Further tests with a higher number of virtual users demonstrated poor performance in some transactions. This was resolved by upgrading the database server hardware.

The final round of tests confirmed all target performance parameters.

Results

With help of WAPT Pro and additional test design services from SoftLogica the client found and fixed several issues, including the limitative server setting for the number of connections and the database server performance shortage. After addressing the problems the system was certified for use in the day-to-day operations inside the client bank.