Why is it necessary to plan a load test before releasing your application ?
Load-testing is the process of testing a system or application under heavy load. Load testing is important to ensure that the system or application will be able to handle the load during the peak hours. Load testing is essential before placing your application into production.
With the rise of the cloud, application architecture is becoming more complex and some micro-services are beyond your control. That’s why testing your application before making it available to your end users is more important than ever. The main rule is that you should test your application as much as possible.
For a successful load-testing campaign, you need to prepare for several things :
- How many users you are planning to have on your application.
- Defining the different user-journeys to test.
- The expected average session duration for calculating the pacing
- Define the list of protection solutions against robots to be deactivated during the test (Captcha, WAF, …)
- Deploy an observability solution like Dynatrace, to get an overview of your application’s health (14-day demo available).
Now that we are done with the load test plan, it’s time to choose a load test solution. At Linkbynet we are working with Octoperf, a French startup that offers a SaaS load testing solution on top of JMeter. This allows us to stay focused on scripting scenarios using their user-friendly interface, and without having to manage the injector pool to simulate traffic.
The main cloud providers are available in Octoperf, so you can have the most relevant test possible. Indeed, if your application is hosted by AWS, you should choose another provider to simulate the traffic, as the cloud provider will optimize the network route and use its internal network. This behavior will impact the result of your test, with a performance that does not reflect reality.
Now that we are done with preparing a load test campaign, why is it crucial to test your application ?
The first reason is to make sure your infrastructure is properly sized. An undersized infrastructure will impact the user experience and will be unavailable during peak traffic. A load test also allows us to detect an oversized infrastructure, and helps us improving auto-scaling rules, for example to reduce the cost of the infrastructure.
The second reason is to assess whether all your third-party services remain available during a traffic peak. Imagine your website fully optimized but with a payment or delivery provider unable to handle a traffic peak…
The last reason is to optimize the static files of your application. Often we are confronted with un-minified CSS & Javascript, or a 20mb Hero image… All this has an impact on the performance of your application and on the cost in terms of business and bandwidth.
