Orchestrating the Resilience of Cloud Microservices Using Task-Based Reliability and Dynamic Costing

In recent years, cloud computing has grown in popularity as enterprises realize the technical and economic benefits of managing resources using an on-demand model. With this growth come user expectations of quality of service, availability and reliability. As the trend in cloud systems architecture moves from monolithic systems to microservices, opportunities exist to develop granular solutions for improving the reliability of the overall system. In this paper we propose an approach to the resilience of microservices-based cloud systems using task-based reliability, quantifying of the reliability requirement with respect to the marginal cost of additional redundancy and the cost of failure. Our approach enables the cloud orchestrator to manage replication of microservice containers to reach an optimal, cost-effective level of redundancy. Empirical testing showed our approach achieved a lower overall cost in comparison with a fixed-reliability approach under an exponential failure function.