TEStore: exploiting thermal and energy storage to cut the electricity bill for datacenter cooling
Proceedings of the 8th International Conference on Network and Service Management
ACM SIGMETRICS Performance Evaluation Review - Special issue on the 31st international symposium on computer performance, modeling, measurements and evaluation (IFIPWG 7.3 Performance 2013)
| Hi-index | 0.00 |
Minimizing the energy consumption of data centers has been researched extensively. However, much less attention is given to a related but different research topic: minimizing the electricity bill of a network of data centers by leveraging different electricity prices in different geographical locations to distribute workloads among those locations. Initial solutions to this problem are oversimplified with an unrealistic assumption that the huge power demands of data centers have no impact on electricity prices. As a result, they cannot be applied to cloud-scale Internet data centers that are expected to grow rapidly in the near future and can draw tens to hundreds of megawatts of power at peak. In addition, existing solutions focus only on server power consumption without considering cooling systems and networking devices, which account for up to 50% of the power consumption of a data center. In this paper, we propose a novel electricity bill capping algorithm that not only minimizes the electricity cost, but also enforces a cost budget on the monthly bill for cloud-scale data centers. Our solution first explicitly models the impacts of the power demands induced by cloud-scale data centers on electricity prices and the power consumption of cooling and networking in the minimization of electricity bill. In the second step, if the electricity cost exceeds a desired monthly budget due to unexpectedly high workloads, our solution guarantees the quality of service for premium customers and trades off the request throughput of ordinary customers. We formulate electricity bill capping as two related constrained optimization problems and propose efficient algorithms based on mixed integer programming. Extensive results show that our solution outperforms the state-of-the-art solutions by having lower electricity bills and achieves desired bill capping with maximized request throughput.