Process-level power estimation in VM-based systems

Abstract

Power estimation of software processes provides critical indicators to drive scheduling or power capping heuristics. State-of-the-art solutions can perform coarse-grained power estimation in virtualized environments, typically treating virtual machines (VMs) as a black box. Yet, VM-based systems are nowadays commonly used to host multiple applications for cost savings and better use of energy by sharing common resources and assets.In this paper, we propose a fine-grained monitoring middleware providing real-time and accurate power estimation of software processes running at any level of virtualization in a system. In particular, our solution automatically learns an application-agnostic power model, which can be used to estimate the power consumption of applications.Our middleware implementation, named BITWATTS, builds on a distributed actor implementation to collect process usage and infer fine-grained power consumption without imposing any hardware investment (e.g., power meters). BITWATTS instances use high-throughput communication channels to spread the power consumption across the VM levels and between machines. Our experiments, based on CPU-and memory-intensive benchmarks running on different hardware setups, demonstrate that BITWATTS scales both in number of monitored processes and virtualization levels. This non-invasive monitoring solution therefore paves the way for scalable energy accounting that takes into account the dynamic nature of virtualized environments.

An intelligent power consumption model for virtual machines under CPU-intensive workload in cloud environment

“…Thus traditional power models for PMs may not be suitable for VMs. Besides, some researches proposed to treat a VM as a process (Colmant et al 2015), which makes it convenient to monitor the amount of resource that a VM occupies at hardware level. However, though a running VM is a kind of process indeed, the VM process differs a lot from a non-VM process with respect to power signature due to virtualization.…”

“…For example, a possible approach to estimate RAM power is to count LLC (Last level Cache) misses. Colmant et al (2015) proposed to sample the count of unhalted-cycles for each application on each physical core. In essence, this process-level power model treats a VM as a process.…”

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

“…With the gathered energy models, it is possible to predict the power consumption of a process, and therefore apply it to estimate the power consumption of the entire VM. This is similar to the work presented by BitWatts [5], which is further capable of estimating the power consumption of a process running within a VM and supports CPU-specific features such as hyperthreading and turbo frequencies.…”

“…In our experiments the actual dynamic costs are measured with a wattmeter as the nodes were used in a single-user mode. For multi-tenant usage a more fine-grained monitoring is required, such as provided by BitWatts [5] that additionally does not require a wattmeter (except for the model building phase).…”

“…An alternative would be to use a software power estimation model that is capable of attributing the dynamic costs to a VM, such as VMeter [3], or Bitwatts [5].…”