Energy-efficient Orchestration Strategies for Function-as-a-Service Platforms

Commercial cloud computing platforms are seamlessly adopted to support many different types of applications, ensuring economic benefits to the application providers. Among the different solutions proposed by cloud providers, Function as a Service (FaaS) platforms are becoming very attractive, since they support lean and agile application packaging and management. However, the energy consumed by the many servers used to support FaaS poses economical and environmental challenges for the Cloud companies. To tackle such challenges, in this paper we design a framework to reduce energy consumption of FaaS platforms by consolidating container creation through active monitoring of the platform status. We implemented our solution on the OpenWhisk platform and carried out a performance evaluation using the Azure Dataset traces. Our experiments show an energy consumption reduction up to 46% with respect to the default resource orchestration of the OpenWhisk platform.