Opt: An Efficient Optimal Autoscaler for Microservice Applications

Microservices are a popular architecture for cloud-based applications subject to stringent performance requirements. To effectively serve variable workloads, autoscaling allocates computational resources ideally at the lowest possible cost. Although several autoscaling techniques have already been proposed in the literature, they suffer from high computational complexity. Here we propose mu Opt as a computationally efficient model-based autoscaler for microservices. By solving a nonlinear optimization problem that embeds a layered queueing network (LQN) model, mu Opt computes optimal configurations maximizing performance while minimizing allocated resources. We validate mu Opt on a benchmark microservice application, reporting fast solution times (similar to 10(-1) s) that enable prompt reactions to highly variable workloads. Compared to a state-of-the-art autoscaler based on LQN and genetic algorithms, mu Opt achieves higher performance (similar to 6%-8%) with significantly fewer allocated resources (similar to 15%-35%) in the presence of both synthetic and real-world workloads.