Solar Energy-Based Virtual Machine Placement Algorithm for Geo-Distributed Datacenters

The cloud computing market is continually rising due to the digital transformation of business sectors, the adoption of information technology devices and the ease of access to the Internet. However, this rise has numerous challenges, including power consumption, customization, security, and compliance. Traditionally, power is generated using fossil fuels, which run the components of datacenters, such as servers, cooling, storage and network systems. Fossil fuels emit carbon dioxide, heat and other gases, directly impacting the environment. Therefore, cloud service providers (CSPs) seek environment-friendly solutions using renewable energy sources to run the datacenters. On the other hand, the datacenters' servers can host multiple customer requests by deploying virtual machines (VMs). The improper placing of the VMs in the servers of datacenters can lead to more energy consumption and increase the overall cost. Recent studies show that VM placement algorithms use renewable energy sources. However, they dispatch the VMs chronologically without considering the types of VMs and servers. This paper proposes a solar energy-based VM placement (SEVMP) algorithm for geo-distributed datacenters. The proposed algorithm dispatches the VMs to the corresponding servers based on their types (groups) and costs, such as power, brown energy usage and fragmentation. Moreover, it selects an appropriate VM from a set of VMs arrived simultaneously to minimize the energy consumption and the overall cost. The SEVMP is simulated by varying VMs, servers and datacenters. The simulation results of SEVMP are compared with four state-of-the-art algorithms to show its effectiveness in the average energy cost, the total energy cost and the number of accepted VMs.