SLA-Aware Predictive Autoscaling for Containerized Microservices

Abstract: Cloud Elasticity can be used for the provisioning and de-provisioning of cloud resources to match incoming workloads with cloud resources in an automatic manner. Elasticity can be achieved using auto-scaling. In literature, major auto scalers discussed can only handle one microservice but a case of multi-service was discussed in a limited fashion. In this paper, we proposed a proactive auto-scaling algorithm to manage the auto-scaling of multi-microservices-based application. compared to the baseline auto scaler (which is a reactive approach and auto scale containers based on the response time of application) our proposed work improves SLA achievement, reduces request failures, and also quickly reacts during heavy incoming workloads.