Energy-based cost model of containers provisioning on clouds

Abstract

Cloud computing revolutionised the development and execution of distributed applications by providing on-demand access to virtual resources. Containerisation simplifies management and support of the cloud infrastructure and applications. Clouds typically are consumed in a pay-as-you-go pricing model. However, when applied to containerised environments, such traditional models do not consider resource utilisation values, leading to inaccurate estimates. Moreover, these models do not consider energy consumption, a dominant component of the data centre’s total cost of ownership. This paper proposes Energy Price Cloud Containers (EPCC), a cost model based on energy consumption that accounts for containers’ effective resource utilisation. We compare EPCC with AWS Fargate to highlight the benefits of using an energy-based pricing model. Thus, by comparing the cost of an application running using AWS Fargate with the estimated cost of that application in nome, it is possible to identify the benefits of using an energy-based pricing model. The weekly costs estimated when running computational resources at nome vary between US$ 2.31 and US$ 10.59. In contrast, when estimating the same amount of resources on AWS Fargate, the costs vary between US$ 2.71 and US$ 29.94. Nome resulted in a cost reduction of up to 35%.