Authors: Bhor Aniket Madhav, Boddawar Kunal Chandrakant, Walke Yashraj Dinkar, Prof. S. L. Rahane, Prof. M. S. Nalawade, Dr. V. M. Dhede
Abstract: Cloud platforms deliver compute, storage and software as on-demand services, yet the data centres behind them draw large and continuous amounts of electricity for servers, network fabrics and cooling plants. Green cloud computing is the body of techniques that lowers this energy and carbon footprint without breaking the service-level commitments made to users. This survey organises the field into four layers of intervention: efficient hardware and facility design, virtualization with dynamic consolidation of virtual machines, energy- and carbon-aware scheduling of workloads, and integration of renewable supply. For each layer the paper explains the underlying principle, reviews representative studies, and compares their objectives, metrics and limitations. Building on this review, a layered conceptual framework is described in which a monitoring module gathers utilization, power and grid-carbon signals, a prediction module estimates near-term demand, and a decision engine places and migrates workloads on the most suitable servers while idle hosts are switched to low-power states. Evaluation through simulation toolkits or a small private-cloud test bed is discussed together with metrics such as power usage effectiveness, energy per request and SLA violation rate. Open issues, including the energy cost of migration, accurate carbon accounting and the trade-off between consolidation and performance, are identified as directions for further study.
DOI: https://doi.org/10.5281/zenodo.23857172
