Success Story:
Major Car Rental Company

Project Description:

The client had decided to move the functionality for servicing rate requests off its VMS-based mainframe to a client/server configuration. The first tier (input/output) takes an Edifac message that is a variable format character string and performs parsing and transformation into a SOAP/XML format. Tier two takes the SOAP/XML and performs all validation processing and creates the queries that go to the third tier (database). The client selected the Microsoft® platform to deliver this functionality, choosing to expose it for consumption by other systems using Extensible Markup Language (XML) Web services. Tier three is a cluster of three four-processor Compaq DL580s running Oracle and the Microsoft Windows® 2000 Advanced Server operating system with Internet Information Services version 5.0, Application Center 2000, Microsoft SOAP Toolkit version 2.0, and several internally developed custom COM+ components. Windows 2000 Network Load Balancing (NLB) is used to distribute the workload across the three processors.
The challenge was to tune the new system and demonstrate that it produced accurate rates in the timeframe required.

ISI provided an on-site project manger responsible for the testing efforts. The manager coordinated client and contract personnel to develop test plans, test conditions, test data and expected results. This information was entered into an automated testing tool used to facilitate the execution of regression tests. The manager also coordinated analysis to determine causation and suggest corrective actions.

A comprehensive automated regression test bed was developed to validate the integrity of the process of loading the data into the database as well as the functional edits inherent in the system. The overhead involved in setting up the automated system was quickly recovered as the system continuously went through regression testing as an integrated part of the defect resolution and configuration management processes. Additionally predictors were developed to estimate the performance of the production system under various load conditions.

Automated tools used to assist in the testing of the system included WinRunner and LoadRunner