Performance Test Result Analysis - Approach Overview |

Performance Test Result Analysis is the most important and technical part of performance testing rather than scripting and execution. A performance tester requires his actual expertise to carry out the test result analysis phase and conclude the testing. This phase is also important because a performance tester has a list of dependent tasks to perform in this phase like analyzing the graphs, checking the metrics, find-out the bottlenecks, comparing the result with defined NFRs and at last, concluding the test result. A small mistake in the result analysis may cause a big disaster in the live environment. That mistake could be due to a lack of experience. Hence a senior performance tester or manager has to review the performance test results before concluding and out for the client.

Purpose:

In this phase, a performance tester determines the bottlenecks and remediation options at the appropriate level – business, middleware, application, infrastructure, network etc. and highlights them in the test report.

Now, the next question comes how to start performance test result analysis so that an accurate test report can be concluded.

This topic is divided into 3 levels to make it simple and easy to understand. The levels are:

Basic Level describes the common graphs of performance testing.
Intermediate Level shows how to analyse client, server and network side graphs.
Advanced Level describes the different approaches to the analysis of performance test result

Before starting the Performance Test Result Analysis there are some Important Points which has to be kept in mind:

You should eliminate “Think Time” from the graph/stats.
You should eliminate “Pacing” (if the tool is considered) from the graph/stats.
No tool-specific error should occur like the failure of the load generator, memory issue etc.
No network-related issues should occur during the test, like network failure, LGs becoming disconnected from the network, etc.
The test should run entirely for the specified test duration.
The result should be collated successfully at the end of the test.
CPU and Memory utilization percentage should be noted down for the pre-test (at least 1 hour), post-test (at least 1 hour) and during the test.
Use proper granularity, so that peaks/lows can be correctly identified.
Optional but helpful – the tool should have a graph merging option available. The analysis of graphs in separate windows makes the analysis work hectic and time-consuming. To make the task easy tool should have a graph merging option. Using this option, you can merge different graphs and investigate the root cause of bottlenecks.
Do not try to extrapolate the result on the basis of incomplete statistics.