The proposed cloud computing scheduling algorithms demonstrated feasibility of interactions between distributors and one of their heavy use customers in a smart grid environment. Specifically, the proposed algorithms take cues from the dynamic pricing and schedule the jobs/tasks in ways that the energy usage is what distributors are hinted. In addition, a peak threshold can be dynamically assigned such that the energy usage at any given time will not exceed the threshold. The proposed scheduling algorithm proved the feasibility of managing the energy usage of cloud computers in collaboration with the energy distributor.
7 Figures and Tables
Fig 1. (a) 14 tasks Gaussian Elimination Structure (b) 25 tasks Laplace Structure
Table 1 Completion time for HLFET and PPAS under three peak thresholds in Gaussian Elimination Structure.
Fig 2. Load profile of Gaussian 189 tasks DAG scheduled by HLFET algorithm, with high energy price and low energy price generated according to its load profile.
Fig 3. Total energy price for Gaussian DAG, using algorithm HLFET, ETF and PAS for high price, low price and random energy price.
Fig 4. Total completion time for Gaussian DAG ,using algorithm ETF, HLFET and PAS for high price, low price and random energy price.
Fig 5. Total energy price for Laplace DAG, using algorithm ETF, HLFET and PAS for high price, low price and random energy price.
Fig 6. (a)Load profile for one Gaussian DAG with 189 tasks and three threshods. (b) PPAS with threshold of peak load of HLFET. (c) PPAS with threshold of algebra mean of HLFET. (d) PPAS with threshold of geometric mean of HLFET.
Download Full PDF Version (Non-Commercial Use)