Serverless Computing Lifecycle Model for Edge Cloud Deployments

Abstract

Serverless computing is a new cloud computing execution model. It offers little operational and administrative overhead, function activation on-demand, and significantly reduced resource consumption. For that reason, it is studied heavily in recent literature by means of handling incoming traffic, applicability for different tasks, and overall performance. Especially the introduction of states, where functionality is kept in CPU or memory can improve performance significantly. However, an in-depth analysis of the resource demand by among others, CPU and memory usage, latency, and energy consumption is still a blank spot in literature. To this end, we propose a serverless lifecycle model with intermediate states and deploy an image processing test setup in the edge cloud to test it. Furthermore, we measure important resource metrics of all states and state transitions of the model. This helps us to answer the question, whether keeping functions available in specific states of a serverless computing instance can improve performance without massive negative influence on the resource requirements.