An Optimization Approach of Container Startup Times for Time-Sensitive Embedded Systems
Lukas Stahlbock, Jan Weber, Frank Köster · 2022
Containers are a lightweight virtualization method that is widely adopted in Cloud Environments, Internet of Things (IoT) and embedded devices. Apart from virtualization, containers provide an efficient way of building, storing and distributing layered filesystems for applications. This is one reason why containers became state of the art for service development, deployment and administration in IT systems. Although containers are lightweight, the image layering and container creation add overhead to the application startup time causing a cold start problem in serverless computing. Therefore, the overhead must be kept as small as possible, especially for resource-constrained devices in time-sensitive systems. We present a concept to decrease container startup times and perform analysis on a resource constrained device, a Renesas RCar H3. The study shows that using our concept, the average container startup time can be reduced by up to 64% on target hardware. Furthermore, we evaluate how the solution scales on different hardware setups and determine the impact of container configuration parameters on startup times.