Why performance testing is essential for user experience

Performance testing is a crucial part of our software development process at Deltatre. A structured and efficient performance testing strategy allows our teams to catch any performance issues that could affect user experience before the application goes live.

Methodologies can vary between companies; however, performance testing focuses on three core areas: speed, scalability and stability.

Types of performance test

There are various performance tests which help to determine how an application will perform once it goes live. Here are some of the most common ones we carry out at Deltatre:

Capacity test

Determines how many users and/or transactions the system can support while continuing to meet performance requirements.

Load test

Looks at how the system handles increasing levels of anticipated loads on the system. This test helps to find any bottlenecks in the system.

Stress test

Tests the application's ability to manage peak loads at or beyond the limits of anticipated loads. Stress testing also tests the ability to handle a reduced availability of resources, such as bandwidth, memory or computing capacity.

Soak test

Also known as endurance testing, this test looks at whether the application can continue to meet performance requirements under the expected load over time. Soak testing helps find any resource capacity problems (e.g. memory leaks or thread pools) that could cause issues over time.

Scalability test

Tests the application's ability to grow beyond current efficiency requirements. For example, how does the system perform with more users or data? Can it do this without sacrificing current performance requirements?

Spike test

Looks at how the system responds to sudden increases or decreases of peak loads. Does it respond correctly and then return to a steady state?

Setting objectives

The first step, before testing begins, is to consider why we are performance testing. What are the objectives? For an efficient and streamlined process, we recommend clearly defining and documenting them.

The reasons for performance testing can vary between projects and organizations. Here are some examples:

Performance-related objectives

• Required concurrency level (e.g. peak level of concurrent users accessing the website, transactions, concurrent playback, etc. )
• Acceptable response time during normal and peak loads
• Expected throughput
• Size of the data to operate on

Capability goals

• Understand the performance-related limits of the project or component (the maximum concurrent/simultaneous consumers)
• Verify/define an infrastructure benchmark/configuration based on performance ratings (environment performance data)
• Gather performance figures for the project or component (e.g. response time, latency, rate per second)

The Deltatre approach

Depending on the objectives, performance testing methodology can vary between companies. The performance testing process we've evolved at Deltatre helps our teams ensure that the software we build meets performance requirements.

Planning

Once we know our test objectives for a given project, we start with the planning stage. We prepare and create the test environment and select the tools we will use. The planning stage helps our teams work more efficiently and identify challenges early on.

Design and Implement

We create the new test scenario and set the performance parameters.

Execution and analysis

To execute the test, we set the stress level and then run the test. If the results are unsatisfactory, we will change the stress level and repeat the test.

Next, we'll analyze the performance results. If they fail the test, some modifications are needed.

Once the scenario has been exhausted, we can set up a new one or exit the test.

Example

The table below shows an example of a load test. The target of this run was to load 1000 virtual users (threads) using 15 load agents in 70 seconds and to keep the load for 1200 seconds.

The benefits of performance testing

A bad customer experience is likely to turn customers away. Performance testing ensures that the app is up to scratch and provides a positive customer experience.

Performance testing allows businesses to see if the application meets the performance requirements (e.g. streaming latency within 1000ms or being able to handle 5000 users).

During performance testing, we monitor the speed, stability and accuracy of an application or system in a range of scenarios, from normal to peak loads. Testing helps us find bottlenecks or errors and identify issues before releasing the product.

Performance testing also allows us to improve load capacity, which can help our clients to increase user volumes when they want or need to scale up.

The key is in the planning. Knowing your objectives and using them to plan the test environments will help ensure an efficient process and help you to identify any issues before going live.