Efficient solar provisioning for net-zero Internet-scale distributed networks
Vani Gupta, Prashant J. Shenoy, Ramesh K. Sitaraman · 2018
Internet-scale Distributed Networks (IDNs) are global distributed systems consisting of a network of hundreds of thousands of servers located in hundreds of data centers around the world. A Content Delivery Network (CDN) is an example of an IDN that delivers content globally through a large network of servers. IDNs consume large amounts of energy and their energy requirements are projected to increase significantly in the future. With carbon emissions from data centers increasing every year, energy solutions for data centers powered by renewables are critical for the sustainability of data centers and for the environment. In this paper, we study the benefits of using solar energy to green IDNs. We study the impact of leveraging global data center locations with high solar potential on the number of solar panels needed to power a CDN. We develop optimal algorithms and heuristics to help minimize the number of solar panels provisioned across the CDN, while making it net-zero. We empirically evaluate our algorithms using extensive load traces from Akamai's global CDN and solar data from PVWatts. Overall, with unrestricted load movement, we can reduce the number of solar panels by 36%, 68%, and 82% for net-zero year, month and week respectively. Our results show that our algorithms can significantly reduce the number of solar panels we need to power our CDN, thereby making sustainability of Internet-scale distributed networks more achievable.