Smart Grid-aware scheduling in data centres
2018-02-06T11:19:45Z (GMT) by
© 2016 In several countries the expansion and establishment of renewable energies result in widely scattered and often weather-dependent energy production, decoupled from energy demand. Large, fossil-fuelled power plants are gradually replaced by many small power stations that transform wind, solar and water power into electrical power. This leads to changes in the historically evolved power grid that favours top-down energy distribution from a backbone of large power plants to widespread consumers. Now, with the increase of energy production in lower layers of the grid, there is also a bottom-up flow of the grid infrastructure compromising its stability. In order to locally adapt the energy demand to the production, some countries have started to establish Smart Grids to incentivise customers to consume energy when it is generated. This paper investigates how data centres can benefit from variable energy prices in Smart Grids. In view of their low average utilisation, data centre providers can schedule the workload dependent on the energy price. We consider a scenario for a data centre in Paderborn, Germany, hosting a large share of interruptible and migratable computing jobs. We suggest and compare two scheduling strategies for minimising energy costs. The first one merely uses current values from the Smart Meter to place the jobs, while the other one also estimates the future energy price in the grid based on weather forecasts. In spite of the complexity of the prediction problem and the inaccuracy of the weather data, both strategies perform well and have a strong positive effect on the utilisation of renewable energy and on the reduction of energy costs. This work improves and extends the paper of the same title published on the SustainIT conference (Mäsker et al., 2015). While that paper puts more emphasis on the utilisation of green energy, the new algorithms find a better balance between energy costs and turnaround time. We slightly alter the scenario using a more realistic multi-queue batch system and improve the scheduling algorithms which can be tuned to prioritise turnaround time or green energy utilisation.