Poster: Controlling Quality of Service of Container Networks in a Hyperconverged Platform

This poster characterizes the impact of storage traffic over containerized network workloads in a shared storage hyperconvergence architecture. We develop CONtrol, a control theoretic approach for assuring the performance of container networks in the presence of disk-intensive workloads. CONtrol uses a proportional integral derivative controller to dynamically decide the bandwidth redistribution among storage workloads and container networks. We have implemented CONtrol over a hyperconverged platform with 5 physical servers, and a thorough testing indicates that it can significantly improve the performance of various benchmark applications over a containerized hyperconverged platform.