The case for early performance testing
What’s wrong with traditional performance testing?
In short, traditional performance testing occurs late in the software development lifecycle – only once a functionally complete and stable application is available.
As a result, it misses chances to fix performance issues faster, which typically occur earlier in development – at a point when many components of the application are yet to be developed, and integration is still some way off.
By adding early performance testing into the performance testing mix you can reduce the
overall time, effort and cost required for testing, while increasing the overall performance of software applications on launch.
Early performance testing is not a magic bullet. It both requires proper planning and implementation and a robust testing toolset.
But when implemented correctly, it helps detect performance bottlenecks early, giving software engineers more time to develop fixes – and helping them do so faster.
With clear early visibility of performance characteristics, all project stakeholders gain deeper confidence in the performance of applications at launch.
So, where exactly does traditional performance testing fall short?
Traditional performance testing
Traditional performance testing
Taking place after the completion of full-scale development and system integration testing, traditional performance testing looks for performance bottlenecks by running tests on the complete application – typically across different business scenario.
Developers use performance monitoring and tuning to try and fix bottlenecks and then re-test the application.
It’s a strikingly inefficient approach.
The problem? In the absence
of early stage testing, we typically find:
- Testing can take as much as 30–60% longer as it’s much harder to identify performance bottlenecks when you only have end-to-end performance results to work with. (It takes longer simply to identify which layer or component is causing the issue.)
- Fixing performance issues can cost as much as 20–40% more as they can require major code rewrites.
3. Shipped software can carry an increased risk of performance issues due to:
- Tight project deadlines leaving insufficient time to fix issues discovered at this late stage in development
- Performance testing becoming restricted in scope due to earlier cost overruns – increasing the risk of it generating misleading results and software shipping with unidentified performance issues.
But these downsides can be sidestepped by adding in early performance testing.
Mastek’s Performance Testing Solution
End-to-end performance testing
Early Performance Transaction Report for HPE LoadRunner
Effective early performance testing rests on deploying an effective performance testing framework alongside appropriate performance testing tools.
For example, to provide first-hand analysis of the performance of an application during early performance testing, Mastek has created the Early Performance Transaction Report for HPE LoadRunner utility.
The utility generates single user transaction response reports in rich HTML and Excel format direct from the vuGen log file. This provides visibility of the application’s performance during early performance testing – reducing the time required for performance test scenario and report preparation by 30%.
Performance Testing & Engineering Toolkit
The Early Performance Transaction Report for HPE LoadRunner utility is just one component within Mastek’s Performance Testing & Engineering Toolkit, an end-to-end performance testing solution combining industry standard and purpose-built in-house tools.
It provides a proactive, comprehensive and repeatable end-to-end performance testing solution that is cost effective, reducing time to market while boosting performance-testing productivity.