Figure 1.
This figure has four kinds of bars, which respectively represent the energy consumption of Random-Migration, DAPSO, PS-ABC and PS-ES.
The four approaches are compared in four time points on energy consumption. The trend line are also added to this figure. The amount of energy consumption is measured by KWh (KiloWatt hour).
Figure 2.
This figure consists of four kinds of curves, which respectively represent the failure rate of VM migration events of Random-Migration, StdPSO, PS-ABC and PS-ES.
X axis denotes the number of VM migration requests. Y axia denotes the failure rate in VM migration events. The failure rate is equal to that the failure number of VM migration events is divided by the number of VM migration requests.
Figure 3.
This figure has two kinds of bars, which respectively represent the failure rate on VM migration events in PS-ES with fixed evaporator factor and self-increased evaporator factor.
Figure 4.
This figure has three kinds of curves, which respectively represent the percentage of the incremental energy consumption of DAPSO, PS-ABC and PS-ES with the percentage of load in the hosts increasing.
X axis denotes the percentage of load in the hosts. Y axis denotes the percentage of the incremental energy consumption.
Figure 5.
This figure has three kinds of bars, which respectively represent the percentage of increase on penalty cost due to SLA violation, energy consumption cost and total cost under varying power management policies of On/Off, Single-DSS, Multiple-SS and Single-SSS.
Figure 6.
This example has shown two kinds of live VM migration situations.
In (a), the situation represents that in cloud data centers there exists the fact that sometimes the VMs with the weight are migrated to the hosts with weak computing power in contrast to the VMs with the light, which are migrated to the hosts with strong computing power. In (b), the situation is similar to that in (a).
Figure 7.
In this figure of PS-ES's architecture, the right part represents the set of available physical hosts, where a plenty number of VMs are running.
In the left part, the controller consists of three modules of Monitor, PS-ES and Migration Controller. It implies that the proposed architecture's processing flow.
Figure 8.
The figure is presented to show how to utilize the SA idea to PS-ES.
These two pie charts describe that PS-ES obtains the final target host of a VM by selecting the solutions with a certain probability value.