Managing cloud computing energy demand in the smart grid environment

Xuan Li, University of Rhode Island


This dissertation focuses on cloud computing centers as the power/energy users in a smart grid environment. Since cloud computing centers are huge energy consumers by definition, the collaboration of cloud computing centers is crucial for the smart grid goals. The proposed method augments the way jobs are scheduled and executed in a cloud computing center, which is primarily in response to the smart grid needs. Specifically, the proposed scheduling algorithm is capable of collaborating with the grander smart grid scheme by (1) acknowledging the dynamic pricing changes and (2) limiting the energy usage of the cloud computing center without any hardware or software alterations. By acknowledging the changes in dynamic pricing, provided by the energy producer or distributor, the proposed scheduling algorithm helps a cloud computing center to participate in the effort to shift the peak load, or to shape the energy load profile. Furthermore, the capability to limit the energy usage on command is valuable for managing the entire power grid in a smart way. ^ Specifically, the proposed “smart cloud” method consists of an upper level “smart cloud scheduling” (SCS) algorithm and a lower level “pricing and peak aware scheduling” (PPAS) algorithm. The SCS coordinates the cloud computing users' jobs and requests, the instructions from the cloud computing manager and the dynamic pricing information from the smart grid. The actual executions of jobs are then assigned to the lower level PPAS algorithm with control information generated by the SCS algorithm. ^ The experimental results show that the proposed SCS and PPAS algorithms successfully achieved the smart grid goals while providing adequate cloud computing services to the users. Specifically, we show that via dynamic pricing, the proposed smart cloud will help in the effort of shifting the peak energy demand on the grid. Moreover, the proposed smart cloud can effectively limit the energy usage of the cloud computers via a careful job and task scheduling without any hardware help. In fact, our experimental results show that the equivalent effect can only be achieved in existing cloud computing by physically turning off part of the system. ^ This dissertation demonstrated that a cloud computing center, as a huge energy consumer, can participate in the smart grid environment. The advantage of the proposed scheme is that we deal only with the changes in scheduling algorithms. There is no need for the development of new materials/devices, hardware or software. In other words, the proposed smart cloud is an immediately deployable scheme.^

Subject Area

Engineering, Electronics and Electrical

Recommended Citation

Xuan Li, "Managing cloud computing energy demand in the smart grid environment" (2012). Dissertations and Master's Theses (Campus Access). Paper AAI3503630.