Energy-aware Processor Management for Virtual Machines

Over the past years, virtualization technology has re-gained considerable attention in the design of computer systems. Virtual machines establish a development path for incorporating new functionality (server consolida-tion, transparent migration, secure computing, etc. ) into a system that simultaneously retains compatibility to ex-isting OSes and applications. At the very same time, with the ever increasing power density and dissipation of modern servers, energy management is becoming more and more a key concern in the design of operating sys-tems. Research has proposed several approaches to OS-directed control over a computer's energy consumption, including user-and service-centric management schemes [1,5]. However, all existing approaches assume full con-trol over and full knowledge of the underlying machine-which obviously does not hold true in virtualized en-vironments. Conversely, current virtualization solutions disregard most or all energy-related aspects of the hard-ware platform, effectively limiting power management to the capabilities of the host-OS in hosted solutions and dispelling it completely from the server-oriented hyper-visor solutions. The key point of this poster is the case for energy-aware, two-level processor management in hypervisor-based server systems. We argue that virtual server in-frastructure needs energy management at both layers, within VMs and across VMs. On the one hand, only guest OSes can pursue fine-grain energy management and keep up guest-intrinsic application-or user-specific service demands. On the other hand, only the hypervi-sor and its resource management subsystem can control global, machine-wide energy requirements and condi-tions. Host-level resource management also remains the resort that enforces given energy requirements for mali-cious, defect, or simply energy-unaware guests. We are currently developing a management frame-work for energy-aware processor management in virtual-ized environments. A host-level scheduler subsystem, re-sponsible for processor allocation to VMs, controls each processor's energy consumption by migrating VMs ac-cording to their respective energy consumptions; it re-sorts to preemption in case that migration should not be sufficient. For energy profiling of VMs, the scheduler re-lies on a runtime accounting infrastructure that estimates the energy consumption by instrumenting performance counters, an approach originally introduced in [1]. The infrastructure instruments both hypervisor and subordi-