Service-Based Container Deployment on Kubernetes Using Stable Marriage Problem

Abstract

Container technology is currently the most widely used technology because of its ease of use and the efficiency of resource usage compared to the use of virtual machines. Moreover, the use of container technology is now made easier by the existence of orchestration tools such as Kubernetes, Docker Swarm, and Mesos. Orchestration tools make it easier for container users to deploy services and control resource. One of the main components of orchestration tools is a scheduler whose job is to manage container placement into a host. In this research, we tried to make a custom scheduler on Kubernetes using Stable Marriage Problem to get the best deployment between containers and hosts. Our trial scenario is to match three web services with three different hosts/servers. From the results of the experiment, containers that are deployed by using the SMP-scheduler have advantages in containers that prioritize CPU usage.