PESS-MinA: A Proactive Stochastic Task Allocation Algorithm for FaaS Edge-Cloud environments

摘要

By the advent of FaaS Cloud services and the micro-services programming architecture, designing task allocation algorithms with higher performance has become a crucial task. Motivated by this high-interested challenge, we propose a new allocation algorithm called PESS-MinA based on our novel modular model for FaaS Edge-Cloud environments. In contradiction to widely-used Max-Min and Min-Min algorithms which are both reactive and deterministic, this algorithm is based on stochastic score, and thus provides proactivity considerations. Experiments with Google Cloud Trace dataset show that our algorithm exhibits better performance in both resource load balancing and QoS assurance of FaaS. According to simulations, PESS-MinA decreased the dropped tasks percentage from 2.9% to 0.01%, alongside with a triple balancing score.