virtual machine placement in cloud computing environment

What is it about?

The ever increasing demand for the cloud services requires more data centres. The power consumption in the data centres is a challenging problem for cloud computing, which has not been considered properly by the data centre developer companies. Especially, large data centres struggle with the power cost and the Greenhouse gases production. Hence, employing the power efficient mechanisms are necessary to optimise the mentioned effects. Moreover, virtual machine (VM) placement can be used as an effective method to reduce the power consumption in data centres. In this paper by grouping both virtual and physical machines, and taking into account the maximum absolute deviation during the VM placement, the power consumption as well as the service level agreement (SLA) deviation in data centres are reduced. To this end, the best-fit decreasing algorithm is utilised in the simulation to reduce the power consumption by about 5% compared to the modified best-fit decreasing algorithm, and at the same time, the SLA violation is improved by 6%. Finally, the learning automata are used to a trade-off between power consumption reduction from one side, and SLA violation percentage from the other side.

Featured Image

Why is it important?

In this paper, using the random learning automata (LA), a new approach is proposed for the dynamic placement of VMs in data centres to reduce the power consumption with the minimum violation of SLA requirements. This approach is based on the live migration and turning off the idle nodes. To be more precise, after grouping the VMs and PMs, a PM with minimum power consumption which has the maximum median absolute deviation (MAD) between all the running VMs and the current VM, is selected.

Perspectives