Analysis model for server consolidation of virtualized heterogeneous data centers providing internet services

Server consolidation based on virtualization technology simplifies system administration, reduces the cost of power and physical infrastructure, and improves resource utilizations in today’s service-oriented Internet data centers. How many servers for the underlying physical infrastructure are saved via server consolidation in virtualized data centers is of great interest to the administrators and designers of the data centers. Various workload consolidations differ in saving physical servers for the infrastructure. The impacts caused by virtualization to these concurrent services are fluctuating considerably which may have a great effect on server consolidation. This paper proposes an analytic model for server consolidation in virtualized Internet data centers based on the queuing theory. According to the features of these services’ workloads, this model can provide the supremum number of consolidated physical servers needed to guarantee QoS with same loss probabilities of requests as in dedicated servers. We verify the model via a case study. The experiments results confirm the superior accuracy of our model and show that the virtual machine-based server consolidation saves up to 50% physical infrastructure and improves 50% CPU resource utilization as well as 2.67 times in I/O bandwidth utilization, satisfying required QoS.