Portfolio

  Risk Management - Quality Assurance

One of our first big projects, Risk Management, was notable for frequent changes in the product requirements, especially at the beginning stage. Numerous alterations brought about plenty of bugs, and qualified testing came in handy. All the testers put their hands to it, and soon most bugs were localized. With the detailed bug description, developers coped with debugging smoothly. In a month we had a stable version and proceeded on increasing functionality. From now on, all new versions of the product received regular testing.

Different stages of the project required various testing methods. With changing requirements, the most effective method was manual testing. In addition to localizing bugs, QA engineers also checked usability of the product. Much effort was given to making the interface user-friendly both for novices and experienced users, so that the latter could quickly perform all their usual repeated operations.

Further on, when the development was mainly aimed at adding new functionality, we turned to regression testing. It appeared that the most effective way then was to partially cover the product functionality with automatic regression tests. Among different available tools we chose Automated QA Test Complete. First of all we automated routine tests where we had to try large amounts of input data. As a result, testing became much faster and more reliable, with wider system coverage – we discovered some bugs which could hardly be found manually due to the amount of data to be tested. The tests which could not be automated were covered with text documentation – test cases and check lists. It guaranteed tests reproducibility and full coverage of functional requirements during regression testing.

Another issue was that of the product operating speed. Our application had to process vast amount of information on surveys, and the most vulnerable point was the time for data processing after the increase in the number of respondents. QA department was to emulate surveys for hundreds of respondents and then test the speed of data processing. We decided to create some automatic tests for filling in the survey and ran them repeatedly. Thus we could check functionality (perform regression testing on survey completion) and test the operating speed. In a week developers had all the necessary information. After some optimization of the process and achieving a time-optimal response, the customer decided to increase the number of respondents fivefold. The process having been fully automated, new surveys were created quickly without additional work hours, and the further optimization began smoothly.

The product under testing was a multifunctional one, with a number of applications working on different platforms and developed with use of different technologies (the desktop part on VB.net based on Windows and the web part in PHP running on a Linux server). As the desktop part closely cooperated with the OS and external applications, we had to perform integration testing. The application was tested while running under different versions of the OS, with different sets of updates and different versions of external programs, as well as without the latter. Automatic tests covered the main part of the functionality, so the whole work took only one month and didn't require any more manpower.

This is how we apply different testing tool in different circumstances striving for the most suitable procedures tailored to particular projects.

You can learn more about the project on www.asuret.com (Collective Intelligence).

 

All Case Studies...