Scalability testing of a production Kubernetes cluster
In early spring 2018 Kubernetes went into production at Meltwater and has seen a huge adoption from a lot of engineering teams. Kubernetes went from hosting 150 services at the end of 2018 to more than 800 services 9 months later.
The team providing Kubernetes as a service wanted to know how much more could Kubernetes grow when onboarding remaining teams with the current setup and architecture.
This presentation describes the findings of the team during the scalability testing of a Kubernetes production cluster: what official information is out there, what metrics to use to determine the health of a cluster, how to simulate a production load and stress a scaled Kubernetes cluster as well as the methodology used for testing.