MFS: A Serverless FaaS Simulator

Serverless computing is one of the concepts that has become a trend these days. Serverless computing facilitates the deployment of software on the cloud, fog, edge, and IoT infrastructures. Serverless computing helps developers implement their codebase without worrying about server management hassles. These are the reasons why the concept of serverless computing is presented in the cloud, the edge, the fog, and the Internet of things. Research on serverless computing, on the other hand, requires extensive hardware equipments. It can also be time consuming. Simulation is needed to overcome these problems. MFS is a simulator based on the Apache Open Whisk architecture that seeks to facilitate the evaluation process of serverless computing researches. Compared to similar researches, this simulator supports more features including containers, heterogeneous processing resources (e.g., CPU, GPU, and TPU), different runtimes, and Apache OpenWhisk architecture. It also calculates and reports metrics such as average response time, utilization of different computing resources, rejection ratio, missed ratio, number of containers used, and costs. The mentioned metrics can be calculated in overall and individually for each physical machine. The status of all functions is also reported at the end of the simulation (e.g., completed, missed, and rejected).