Kubernetes Performance and Load Testing Tutorial

Performance testing is critical for any software that goes to production. For an application deployed via a Kubernetes cluster, test to ensure that the cluster scales to meet changes in request volumes.Kubernetes performance testing demands a place in the software development lifecycle for container-based applications. The goal of any type of performance test is to build highly available, scalable and stable software. Load, endurance and stress tests reveal how the system responds in various situations. For example, a web application might handle 100 requests at a time, but what happens when 1000 or more requests come in — will it crash, or serve the incoming requests with a slower response time? The types of performance tests are:

Load test. How the system responds to a sudden increase in requests.
Endurance test. How the system survives a constant, moderate load for longer duration of times. It can also be referred to as a soak test, referring to the long time the software spends under test.
Stress test. How the system responds under a heavy load, with an intent to find out the point at which system is stressed and ceases functioning.

To demonstrate load testing, this tutorial puts a lot of demand on a PHP application running in a Kubernetes cluster. The aim is for the cluster to scale horizontally when incoming requests exceed normal usage patterns.

Install the Metrics Server

The Kubernetes Metrics Server is the crucial component for a load test because it collects resource metrics from Kubernetes nodes and pods. Metrics Server provides APIs, through which Kubernetes queries the pods’ resource use, like CPU percentage, and scales the number of pods deployed to manage the load. There are multiple ways to install it on a Kubernetes cluster.

Helm. Helm charts, collections of files for related Kubernetes resources, are a popular way to set up Metrics Server on a cluster. Run the following command:

helm install metrics-server stable/metrics-server

Helm charts are created and maintained by open source community contributors, hosted on GitHub.

DIY setup. To directly deploy the Metrics Server, use the following command and a YAML file hosted on GitHub:

kubectl apply -f https://github.com/kubernetes-sigs/metrics-server/releases/download/v0.3.6/components.yaml

Once the above command executes successfully, run the following command to verify that Metrics Server was installed:

kubectl get deployment metrics-server -n kube-system

As a cloud service. Google Kubernetes Engine (GKE) and Microsoft’s Azure Kubernetes Service (AKS) ship Metrics Server as a part of the GKE or AKS Kubernetes cluster, respectively, and therefore the user does not have to install it explicitly.Amazon Elastic Kubernetes Service (EKS) from AWS supports Metrics Server. EKS users should set up Metrics Server following the DIY setup method offered above.

Locally. If the Kubernetes cluster runs on Minikube, a tool for local Kubernetes deployment, enable a Metrics Server add-on:

minikube addons enable metrics-server

With the Metrics Server installed and capturing the resource metrics in the Kubernetes cluster, deploy the application on which to run performance tests.