Adaptive container auto-scaling for fluctuating workloads in cloud

Xiaoyue Feng, Sijia Zhang, Tianzhe Jiao, Chaopeng Guo, Jie Song

Abstract: Highlights•Proposes an auto-scaling method resource container in the cloud that adapts to the fluctuating workload.•Designs a resource estimation model that uses a graph neural network to predict resource demands for container scaling.•Designs an improved RL-based algorithm with a dynamic action model for different variations in resource demands.