EFaaS: Energy-efficient Function Orchestration in Serverless Edge Computing

Abstract

Energy in computing is critical as sustainability concerns rise. The serverless edge computing’s energy consumption is significant and growing rapidly. We propose EFaaS, a serverless scheduler that dispatches and caches functions, integrated into an existing serverless platform, reducing the energy cost by jointly considering request dispatching and function caching. We evaluate it using simulations with real-world traces. EFaaS achieves superior performance with $36.5 %$ reduction in energy cost.