Monitoring Container Services at the Network Edge

Recent developments induced by the Internet of Things (IoT) force a paradigm shift to deploy on demand services to a broad range of different computing architectures. Mainly single board computers (SBCs) gained a lot of attraction in recent years, shifting highly available processing power to service consumers and IoT devices. Simultaneously, container virtualization achieved a breakthrough by the well known Docker environment that became a key competitor to the utilization of virtual machines. Nevertheless, spreading containers at small scale needs a proper allocation of resources available on SBCs. As a first step, we developed a multi-architecture framework "PyMon" to monitor different computing architectures with a small footprint. PyMon itself is based on the recent version of "monit" and a "Django" application to collect monitoring data. It is delivered by several Docker images and allows monitoring with a reasonable processing overhead. Our demonstration shows the statistics of either processes or containers running on a cluster of SBCs, where each member is monitored. Furthermore, we evaluated the resource usage of the two key competitors for container cluster management on SBCs, Kubernetes and Docker Swarm, with PyMon.