The power consumption of presently available Inter-net servers and data centers is not proportional to the work they accomplish. The scientific community is attempting to address this problem in a number of ways, for example, by employing dynamic voltage and frequency scaling, selectively switching off idle or underutilized servers, and employing energy-aware task scheduling. Central to these approaches is the accurate estimation of the power consumption of the various subsystems of a server, particularly, the processor. We distinguish between power consumption measurement techniques and power consumption estimation models. The techniques refer to the art of instrumenting a system to measure its actual power consumption whereas the estimation models deal with indirect evidences (such as information pertaining to CPU utilization or events captured by hardware performance counters) to reason about the power consumption of a system under consideration. The paper provides a comprehensive survey of existing or proposed approaches to estimate the power consumption of single-core as well as multicore processors, virtual machines, and an entire server.

Abstract—The adoption of server virtualization and cloud computing has enabled high flexibility of service execution in the Internet. It also promises the efficient use of resources including power. At present, the cloud infrastructure (physical machines and cloud platforms) and the services employing the infrastructure are managed by independent entities. As a result, it is difficult to jointly configure hardware and software resources, which may introduce significant inefficiency of resource utilization. Often infrastructure providers over provision resources to accommodate a growing demand, but the cost of such inefficiency is gradually being felt by both parties. This paper experimentally examines the effect of system configuration (concurrency) on the power consumption and latency of a video hosting server. We find that the usefulness of concurrency is greatly influenced by the interplay of underlying leased resources and by the interaction of virtual machines with these resources. However, the exact nature of this interplay is difficult to quantitatively establish and, therefore, it is not presented to service providers. Our study encourages the scientific community to pay attention to this aspect and to undertake a more rigorous investigation based on practical observations. Index Terms—Concurrency, parallel programs, power consumption, server power consumption, processor power consumption, virtual machines, latency, performance, virtualized environment I.