Databases Reference
In-Depth Information
Setting up the load testing environment
Load testing is normally performed after functional testing is done. Once we
are ready with load testing scripts, the next step is to set up the load testing
environment. The load testing environment should be of one that is in production
such as hardware and deployment configuration. As the purpose of load testing is
to verify if the system passes performance expectation with the pick time load, load
testing and tuning should be done on a production-like environment. All necessary
resource monitoring tools should be installed and configured. Enough space should
be allocated for storing various monitoring logfiles.
JMeter should be set up on separate machines with enough memory and CPU
capacity. If the number of concurrent users is too high, it is recommended to run
load tests from multiple machines. In general, we can run load tests with around 300
concurrent users from a machine with 2 GB RAM allocated to the JMeter application.
Conducting load tests
Load testing is an iterative exercise. It is highly integrated with the resource
monitoring and performance tuning exercise. The following diagram explains
the iterative process of conducting load tests:
3. Performance
Tuning
1. Test Execution
2. Result Analysis
As shown in the diagram, the first step is to execute the load test and monitor
resources. The second step is to analyze the load test results and the resource
monitoring data. Based on the analysis, necessary changes in the environment,
source code or load testing scripts should be produced and then again the same
cycle should be followed until expected performance targets are achieved.
As mentioned earlier, JMeter provides good GUI-based reporting plugins. Using many
reporting plugins can slow down JMeter. Hence, it is recommended to use only those
reports that are necessary. Most of the time, we would like to capture average response
time, response time of 90 percent requests, throughput, error percentage, and so on.
These results can be captured using the aggregate report of JMeter.
