INCENTIVE-SCHEDULING ALGORITHMS TO PROVIDE GREEN COMPUTATIONAL DATA CENTER

The increased of computational loads in grid servers are dissipating more heat and eventually amplifies the cooling demand in Data Center (DC), and at the same time lead to more submitted jobs missing their job completion deadlines. Accordingly, power supply for cooling such systems in a DC is up until 40% to 50% of the total power consumption. To mitigate the cost of electricity, Gas District Cooling (GDC) may be deployed to provide electricity and chilled water to DC facilities with relatively low running cost. Using GDC-DC control model, a GDC can be made to channel wasted energy to cool a DC, effectively reducing CO2 emission. However, it has not been shown how a job scheduler can exploit the GDC-DC model, especially which runs on small-scale DCs. Hence, this study proposed a cooling-efficient job scheduling algorithm for small-scale DC (cooling energy consumption saving can be effective at any conditions e.g. 25 job controlled resources used in this research) that could not accommodate a full-scale GDC-DC model (cooling energy consumption saving can only be effective at specific conditions; needed 6500 job controlled resources). A control strategy was deployed such that the computational job schedulers can reduce the cooling energy consumption in DC by shifting heavy jobs to be executed at night time but light jobs at day time. The experiment results significantly show that the proposed method able to reduce the cooling electricity consumption by 23%. This however may lead to more missed deadlines and unfortunately the controlled resource grid scheduling does not provide compensation to the resource users upon deadlines that missed. Furthermore, the absence of compensation for missed deadlines may dissuade users from submitting jobs to the grid. To solve this issue, green incentive based scheduling algorithms were devised to significantly save cooling electricity consumption cost in DC and also be able to compensate users for their submitted jobs that missed the job completion deadline(s).