Figure 3 IP risk during off-peak hours using Risk Min In terms of energy efficiency, Fig. 5 and Fig. 6 show that proactive mode of EE Max policies provides lower energy ef-
Related Figures (10)
In our experiments, we evaluate 3 metrics. a) Availability: the amount of time in which the VM runs the services and they are available to the client. We consider that the VM is not available when it (or the host running it) has crashed, or when the VM is being restarted, or migrated. b) IP risk level calculated using the IP failure risk model. c) IP energy effi- ciency, measured as a percentage. 100% means that the nodes that are switched on are using all their CPUs. Nodes without workload are switched off, and they will automatically switch on when the running nodes do not have enough resources for the incoming requests. 0% would mean that all the nodes are switched on and there are no running tasks. Fig. 2: IP risk during peak hours using EE Max Fig. 1: IP risk during off-peak hours using EE Max Fig. 5: Energy efficiency during off-peak hours using EE Max Fig. 4: IP risk during peak hours using Risk Min Fig. 6: Energy efficiency during peak hours using EE Max Fig. 7: Energy efficiency during off-peak hours using Risk Min Fig. 9: Availability of VMs using Risk Min Fig. 8: Energy efficiency during peak hours using Risk Min Fig. 10: Availability of VMs using EE Max