Can Software Containerisation Fit The Car On-Board Systems ?

As the automotive industry evolves towards inter-connected and intelligent vehicles, the integration of complex electronic and software components has become paramount. However, this increased complexity brings new challenges in ensuring system safety, security, and life-cycle management. In this article, we focus on virtualisation (virt.), particularly containerisation (cont.), as a solution to mitigate integration stress in multi-node environments. We present a detailed review of the automotive constraints and ecosystem, along with the selection criteria for virt. technologies, justifying the scope of our study. Our main contribution is a two-phased evaluation of cont. tools. Firstly, we assess popular single-node container engines (Docker, Containerd, and Linux Containers LXC) based on CPU, RAM, and file I/O overhead over multiple hardware configurations. Secondly, we evaluate their multi-node scalability. The results show that Docker performs in average better than the other solutions for automotive on-board architectures.