No abstract

No abstract

No abstract

A key challenge for supporting elastic behaviour in cloud systems is performant automated (de-)provisioning and scheduling of computing resources. Containers are rapidly replacing Virtual Machines (VMs) as the compute instance of choice in many cloud deployments, due to lower overhead of starting up and terminating containers in comparison to VMs. Performance overheads associated with deploying, terminating and maintaining a container can be significant. We analyse performance of the Kubernetes system and develop a Petri net-based model of resource management within this system. Our model is characterised using data from a Kubernetes deployment, and can be used as a basis to support capacity planning and design of scaleable applications that make use of Kubernetes.

No abstract