This paper explores the evolution of application architectures in the context of edge computing, tracing a journey from monolith to microservices and finally arriving at a microkernel-based approach. The initial challenge involved containerizing monolithic applications for deployment on low-powered edge devices, where resource constraints and maintenance issues pose significant hurdles. The transition to microservices within application boundaries addressed some concerns but introduced new challenges related to increased resource consumption. Recognizing the need for further optimization, the paper discusses the adoption of a microkernel architecture as a strategic solution. This architecture allows on-demand loading of required plugins, enabling the dynamic activation of specific features. The in-process nature of these plugins fosters faster execution and communication, proving crucial for efficient resource utilization on devices with limited processing power. The paper details the microkernel architecture’s implementation, emphasizing the loading of plugins on demand and the resulting benefits in terms of memory footprint and execution speed. Results from practical applications illustrate significant improvements in resource efficiency compared to previous architectures. The microkernel approach not only addresses the challenges posed by microservices but also aligns well with the unique demands of edge computing.