Container Restart Reduction Technique in Kubernetes Using Memory Oversubscription
Taeshin Kang, Heonchang Yu, Eunyoung Lee · 2023
This paper proposes a technique to mitigate container restarts in a memory oversubscription environment based on Kubernetes. The proposed technique involves identifying containers that are likely to request memory allocation on nodes experiencing high memory usage and temporarily pausing these containers. By significantly reducing the CPU usage of containers, the similar effect as pausing them can be achieved. The suspension of the identified containers is released as soon as it is determined that the corresponding node’s memory usage has been reduced. The proposed method reduces the average number of container restarts by a maximum of 58% and reduces the total run time by an average of 7% and a maximum of 13% for high memory volatile workflows in Kubernetes environments.