The impact of management operations on the virtualized datacenter

Abstract

Virtualization has the potential to dramatically reduce the total cost of ownership of datacenters and increase the flexibility of deployments for general-purpose workloads. If present trends continue, the datacenter of the future will be largely virtualized. The base platform in such a datacenter will consist of physical hosts that run hypervisors, and workloads will run within virtual machines on these platforms. From a system management perspective, the virtualized environment enables a number of new workflows in the datacenter. These workflows involve operations on the physical hosts themselves, such as upgrading the hypervisor, as well as operations on the virtual machines, such as reconfiguration or reverting from snapshots. While traditional datacenter design has focused on the cost vs. capability tradeoffs for the end-user applications running in the datacenter, we argue that the management workload from these workflows must be factored into the design of the virtualized datacenter. In this paper, we examine data from real-world virtualized deployments to characterize common management workflows and assess their impact on resource usage in the datacenter. We show that while many end-user applications are fairly light on I/O requirements, the management workload has considerable network and disk I/O requirements. We show that the management workload scales with the increasing compute power in the datacenter. Finally, we discuss the implications of this management workload for the datacenter.