The effect of Bandwidth Allocation on power efficiency in cloud data centers

Cloud computing is an emerging model in the field of computing. It has been the most studied model in recent years due to the huge amount of energy consumed in the cloud data centers. A great deal of studies focused on investigating the effect of virtualization management on improving energy efficiency of the cloud data centers. However, most studies only considered task resource requirements of processor, memory, and storage without considering bandwidth requirements and its effects on energy consumption. In this paper, two directions that tackle the bandwidth requirements are presented. In the first direction, an algorithm for reallocating the bandwidth for the virtual machines hosted on the same physical machine is proposed. This algorithm, called Fair Bandwidth Reallocation (FBR), reallocates the bandwidth amounts between two virtual machines based on requirements of the jobs executed on these VMs at a specific period of time. The core idea of FBR is to maximize the utilization of the bandwidth of the PM. One virtual machine (The Granter) provides a specific amount of its bandwidth to another virtual machine (The Grantee) when it is idle, and then regains this amount after a period of time. In the second direction, another algorithm is proposed to dynamically allocate bandwidth to the virtual machines that are hosted on the physical machine. This algorithm, called Required Bandwidth Allocation (RBA), treats bandwidth as a shared resource and allocates only the amount of bandwidth required to the VM at that specific time. The core idea here is to monitor the bandwidth of the virtual machines periodically. RBA reallocates the bandwidth among these virtual machines according to their actual required bandwidth.