Research On Energy Management Of Geo-distributed Cloud Data Centers Optimized Cost And Carbon Emission

With the continuous increasing of internet services,the number of cloud computing data centers increases,and the scale is increasing,resulting in the rapid increase of power consumption and power cost.The demand of energy for power is huge,but the energy resource is mainly derived from fossil fuels,which makes the carbon emissions rise sharply.Based on the grid price fluctuates and the green energy can not powered continuously,this paper proposed the green scheduling of geo-distributed cloud data centers.The request come from geo-distributed data center has multi-dimension resource.The data center powerd by green energy and grid directly and indirectly by energy storage devices charging and discharging.It is important for studying the energy consumption of data centers and protecting the environment.This paper mainly studies the energy management of geo-distributed cloud data center based on minimum cost and the energy management of geo-distributed cloud data center based on minimum carbon emission under the budget constraint respectively.For the energy management of geo-distributed cloud data centers based on minimum cost,the model of minimum total cost is established and propose a design strategy of sliding time window based on the electricity price fluctuation of subsequent ? time slots.This paper analyzed how to distribute as many requests as possible to data center with low electricity price supplied by grid,for the incoming multi-dimensional resources requests deployed to as few servers as possible.In resource scheduling,we mainly use the best-fit descending bin-packing algorithm that deploy as many requests as possible to the most appropriate server,resulting in less energy consumption.For the goal of the minimum cost,combined with the current market electricity prices fluctuations and the conditions of workload,compared with no use energy storage devices or use energy storage devices charging and discharging,and different strategies of charging and discharging.The real cloud data centers servers running log are used to simulate the experiment,its show that the design strategy of sliding time window reduces the cost of electricity greatly.For the energy management of geo-distributed cloud data center based on minimize carbon emission under the budget constraint.The model of minimum carbon emission within budget constraints is established.According to the characteristics of intermittent power supply of green energy,this paper proposed a strategy by using sliding time window based on the subsequent ? time slots with green energy and budget constraints.This paper analyzed how to reasonably schedule green energy to charge and discharge for energy storage devices,which makes the data center have the least carbon emission under the budget constraint.When schedule resource,allocated the budget by time slots reasonably.Always distributed as many multi-dimensional resource requests as possible to data centers with more green energy and low price,as much as possible to reduce carbon emissions under the budget.The experiment uses the same running log of the real cloud data centers,combined with the current market electricity prices fluctuations,uninterrupted power supply of green energy and the conditions of workload,The charging and discharging strategy of the sliding time window based on the subsequent time slots reduces the carbon emission significantly.