A Cost-Variant Renewable Energy-Based Scheduling Algorithm for Cloud Computing

The global growth of cloud computing services is rising abruptly due to a large variety of services like computing, storage, network, etc. It expresses cloud service providers (CSPs) for better usage of existing datacenter resources, increasing agility, and reducing the need for unanticipated datacenter growth. These datacenters use a lot of energy generated from fossil fuels (i.e., non-renewable energy (NRE) sources) and omit a lot of nitrous oxide and carbon dioxide, which cause the greenhouse effect and are harmful to the environment. Moreover, NRE sources are limited in supply and cannot be sustained over a long period. As a circumstance, CSPs are moving towards renewable energy (RE) sources, such as solar, wind, hydro, and biomass, to decarbonize datacenters even though these resources are not available round the clock. Therefore, recent studies focus on using both RE and NRE sources to avoid any interruption of the datacenter services. However, these studies consider the equal cost for all the RE sources and do not consider the categorization among user requests (URs). This paper considers the different costs for RE sources and two categories of URs, namely critical and non-critical, and introduces a cost-variant RE-based scheduling (CRES) algorithm for cloud computing. Here, the critical UR does not depend on the RE resources due to the unpredictability of RE sources. On the other hand, the non-critical UR can be accommodated by both RE and NRE resources. We simulate the proposed algorithm by considering 20 to 100 URs and 5 to 25 datacenters and compare the performance with the future-aware best fit (FABEF) and highest available renewable first (HAREF) algorithms in terms of cost and usage count of RE resources to show its usefulness.