QoS for best‐effort batch jobs in container‐based cloud

The resource orchestrators in the cloud provide different QoS classes. Existing studies have mainly focused on guarantee for high-priority class jobs, but yet to consider delicate control of low-priority jobs cloud. In this paper, we propose differentiated shares that weighted CPU scheduling batch container-based To implement shares, extend reservation interfaces provided by Kubernetes and suggest an algorithm maps management attributes into those Linux. proposed can avoid worsening interference with containers exploiting hierarchical sharing task groups addition, a node scoring policy efficient inter-node job consideration shares. performance measurement results show best-effort without (less than 3% respect application-level performance).