At a glance: Azure Load Tests Service
Quality metrics are important and need to be measured when we build and deploy a new system version. Resilience, response time, scalability, and application performance are not easy to test when creating a cloud solution.
Today’s post talks about how you can run a performance test on top of Azure using Azure Load Testing. This service provided by Microsoft gives us the possibility to run large performance, scalability or application quality tests in a controlled and easy manner.
With the Azure Load Testing service, a developer or tester can configure and run a load test in just a few minutes, collect the output, and identify the system’s bottleneck.
Azure Load test components
4 main components are built around Azure load and performance capability:
(1) Azure Monitor — used to collect information from Azure services
(2) Azure Application Insights — used to collect application data and to provide an easy way to display and track application metrics
(3) Azure Container Insights — used to collect containers data and integrate the output with other monitoring systems (Azure Monitor, Azure Application Insights)
(4) Azure Load Testing — used to run the performance and stress tests and orchestrate the load test(s)
One of the nice things provided by the new service is the dashboards that are consolidating in one location all the metrics from client-side and server-side, including HTTP responses, DB load and reads, container resource consumption, and so on.
As expected, there is full integration with Azure Pipelines and GitHub Actions. We can specify a performance baseline and trigger a build using the pipelines when the performance criteria are not matched.
How to run a load test?
The loads’ tests are built using Apache JMeter scripts. We can easily reuse the JMX files we are using for other kinds of difficulties or on-premises systems if we migrate from on-premises. The JMX files can be uploaded directly to the Portal or from the repository. Another way to run the load tests is based on the classical ‘Test method’.
Azure Load Testing offers us the ability to integrate with CI/CD workflow. We can trigger the Load Testing step after we build and deploy the application into the testing environment.
This step loads the Test Plan configuration and calls the Azure Load Testing Service. Once the load tests are run, the test results, together with all collected metrics, are pushed back to the CI/CD workflow and depending on the test results, the workflow step will pass or fail. Azure Load Testing Dashboards are available after each run to be analyzed by the testing or development team. Additional actions can be registered based on the test results.
The pricing model has 3 main components:
(1) Resources that are used by the system when you run the load test
(2) Resourced used by the Load Test to create the load and run the test engine
(3) Additional usage of Virtual Users Hours
Should we consider reviewing and integrating this service? YES. In the long term, the impact of Azure Load Testing on how you monitor the quality metrics is enormous. Try to do the integration in small steps and ensure that you have the business and quality metrics of the system before you start running to run performance and load tests. Start from business drivers and expectations before jumping to run the stress and quality metric tests.