Improving Renewable Energy Usage in Data Centers with Distributed UPS Using Batch Job Scheduling

Please enable javascript in your browser.

Haghshenas, Kawsar | 2014

567 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 46262 (19)
University: Sharif University of Technology
Department: Computer Engineering
Advisor(s): Goudarzi, Maziar
Abstract:
In recent years, the demands come to data centers have increased significantly, so huge amount of energy consumption, cost of this usage and environmental effects of data centers have become serious issues. Accordingly using solar energy to provide some portion of a data center’s energy usage can reduce environmental effects efficiently. Also smarter use of this energy could decrease overall cost of energy in data center. The demands that come to a data center are categorized to critical and non-critical workloads. Running of Non-critical workload (batch job) can be moved to another time according to its deadline. This work suggests a solution that schedules batch jobs considering solar energy, cooling subsystem, rate structure and Uninterruptible power supply to reduce cost of energy. We evaluated our solution using simulations and Real Benchmarks. The results show our solution can reduce cost of energy up to 50% compared to existing approaches
Keywords:
Data Center ; Solar Energy ; Job Scheduling ; Uninterruptible Power Supplies ; Energy Optimization ; Energy Consumption ; Batch Job ; Free Cooling

No TOC