How Much Does a VM Cost? Energy-Proportional Accounting in VM-Based Environments

Abstract

Abstract-The costs of current data centers are mostly driven by their energy consumption (specifically by the air conditioning, computing and networking infrastructure). Yet, current pricing models are usually static and rarely consider the facilities' energy consumption per user. The challenge is to provide a fair and predictable model to attribute the overall energy costs per virtual machine (VM). Current pay-as-you-go models of Cloud providers allow users to easily know how much their computing will cost. However, this model is not fully transparent as to where the costs come from (e.g., energy). In this paper we introduce EPAVE, a model for Energy-Proportional Accounting in VMbased Environments. EPAVE allows transparent, reproducible and predictive cost calculation for users and for Cloud providers. We show these characteristics of EPAVE by a number of use cases in heterogeneous data centers and discuss the applicability of EPAVE.

Estimating Energy Consumption of Cloud, Fog, and Edge Computing Infrastructures

“…As Figure 4 shows, our model divides energy consumption of an ICT equipment into static and dynamic parts [32]. The static energy consumption is the energy consumption without considering any workload (i.e., resources are idle).…”

“…As Figure 4 shows, static energy part of a Cloud-related architecture includes static energy of PMs and switches. For a PM, the idle power consumption corresponds to the power consumed by the server when powered on but not running any task [32]. Similarly, the energy consumption of loadindependent part of a switch is considered as its idle power consumption.…”

Energy efficient computing, clusters, grids and clouds: A taxonomy and survey

“…The privileged hypervisor have control over virtualised server -they can get coarse-grained energy consumption of the server. However, fine-grained VM level energy consumption cannot be measured in this way [44], as the capability of server is multiplexed across several VMs, whose real energy consumption is decided by the characteristics of running application. Moreover, the virtualization layer makes it challenging to isolate application's energy consumption from server's total energy consumption.…”

“…Assuming a constant power use for a server that is switched on with no VMs, the first VM is going to be least energy efficient -the baseline energy use gradually spreads across all VMs. Therefore new VM level energy consumption models are needed to measure accurate levels of total energy consumption across virtualized platform [44]. In our previous work [45], we proposed an energy consumption model for virtualized systems, to decide which type of host could run a VM more energy efficiently.…”

A CO2 Emissions Accounting Framework with Market-based Incentives for Cloud Infrastructures

“…The energy consumption of their virtual machine depends on the resource utilization (CPU, memory, disk, network) and on the infrastructure power management (cooling cost, resource allocation management, etc.) (Kurpicz et al, 2016). The CO 2 emissions depends on the energy consumption and on the electricity mix (Wajid et al, 2015).…”