PERFORMANCE TESTING FOR THE SELF SERVICE APPLICATIONS - (METHODOLOGY,ISSUES, BEST PRACTICES )
The entire lifecycle and testing approach which should be adopted for conducting a performance testing approach for a newly implemented application which is highly critical in terms of determining the system behavior and performance for handling the user load and helps us determining the effectiveness of the application. It helps the team to baseline their user load and simulate the production scenarios in the test environment and fine tune the performance related glitch.Once the issues are resolved it helps us simulate the production load and thus enhance the overall performance of the system when it goes live.
Key Activities
Ø Mining data from historical information
Ø Interviewing key stakeholders for baselining the requirements for testing.
Ø Gather performance testing SLA's
Ø Understand percentage of expected invalid requests
Ø Prioritize test requirements based on criticality and time
Ø Document performance testing requirements
Ø Review and signoff on Test requirements
Ø Performance test team to drive the requirements collection process
Ø System Architects, Development team and other technical experts
Ø Business Analysts, Subject matter expert, End Users and other functional experts to provide required inputs
Issues/ Challenges (Performance testing) :
Test Scripting Phase :
1. Too many correlations make the script look complex. Due to the dynamic variables for the fields in the adobe form, the script gets many correlated values in it.
2. The effective dates should always be maintained as +/- 30 days from the current date. The date was hard coded and at times it requires validations to check the date when there is a failure in the script.
3. For some complex and long forms if there are many fields in the script it makes it complex while recording and prone to errors while execution
Test Execution Phase
§ Performance testing environment and Production Environment will not be similar.
§ Frequent functional change activities planned at regular Intervals will impact the test scripts due to changes in the application functionality.
§ Coordination between different teams for Test data preparation and execution
§ For some tests with high user load, few links disabled after a good amount of transactions. (ESS link, Pay Change, Address info, etc...)
§ Executions fail due to the acceptance of minimal number of users by the servers. After certain number of transactions the error rate ramps up at a faster rate.
§ Frequent locking of ESS test id's which caused failure. Sometimes, these id's are forcefully released from the backend to go with the test.
§ For few tests, we get time out errors. When there is a server restart or the cache is cleared, the first few transactions might take a very long time. These errors can be taken care by increasing the Request time out.
§ For some tests, if the employees reporting changes in between the cycles of test with different reporting managers, the original scripts approval workflow scripts fails. Care should be maintained not to change any such data for the test dataset.
§ For few tests, there are initialization errors. This must be due to the insufficient space on the hard disk.
§ Due to non-availability of required network infrastructure, there can be minor and major slippages in testing schedules.
Solution Approach and Best Practice in Performance Testing
Ø Centralized performance test team that takes ownership of testing each release prior to production deployment
Ø Worked closely with the Functional team for KT, Requirement understanding and Test Data Creation
Ø Identifying the Performance Critical scenarios across the applications.
Ø Automating the test scripts for the scenarios using automation tools.
Ø Test execution Results analyzed and Performance Bottlenecks identified
Ø Analyze the impact of network bandwidth and Latency on application performance
Ø Performance test reports should be presented to the client in a customized, simple reports summarizing the facts instead of the reporting the complex tool generated system report which will not be user-friendly.
Ø Data setup should be made easy through automated scripts which had significant time savings too.
Ø Improved coordination by involving multiple technical and business teams during test execution and results analysis
Ø Maximum knowledge retention through respective Centre of Excellence.
Ø Run tests in exclusive test window not impacting other applications as Environment is normally shared.
Ø Run Health Check tests to check the availability of the applications before running a load test
Ø Setting up Performance Test lab for network simulation.
Ø Process documented knowledge transfer for future testing which can be baselined and referenced .
