Energy-aware elastic scaling algorithm for microservices in Kubernetes clouds

Current elastic scaling algorithms are limited to the perspective of containers, and applications running within containers are treated as monolithic applications when designing scheduling algorithms. In addition, the default scaling mechanisms in Kubernetes fail to effectively distinguish and manage resource consumption of idle containers, leading to resource waste and degraded system performance. Therefore, this paper proposes an energy efficiency model based on Service Level Agreement (SLA) and an energy-aware elastic scaling algorithm based on SLA to reduce the energy consumption of microservices deployed in Kubernetes. The proposed algorithm optimizes the feedback control method in container provisioning by releasing excess container resources, and periodically runs the feedforward control algorithm and the feedback control algorithm to effectively deal with the dynamic changes of the workload. The experimental results show that the energy consumption of Kubernetes clusters in a cloud environment can be reduced by 15.34% compared with the default elastic scaling algorithms in Kubernetes and the latest algorithms.