In order to improve locality aspects, new Cloud-related architectures such as Edge Computing have been proposed. Despite the growing popularity of these architectures, their energy consumption has not well investigated yet. To move forward on a critical question, we first introduce taxonomy different architectures. From this taxonomy, then present an model evaluate consumption. Unlike previous proposals, our comprises full computing facilities, including cooling systems, and network devices...

The global energy demand for digital activities is constantly growing. Computing nodes and cloud services are at the heart of these activities. Understanding their consumption an important step towards reducing it. On one hand, physical power meters very accurate in measuring but they expensive, difficult to deploy on a large scale, not able provide measurements service level. other models vendor-specific internal interfaces already available or can be implemented existing systems. Plenty...

While an extensive set of research project deals with the saving power problem electronic devices powered by electric battery, few have interest in large scale distributed systems permanently plugged wall socket. However, a rapid study shows that each computer, member system platform, consume substantial quantity especially when those resources are idle. Today, given number processing involved computing infrastructure, we convinced can save lot applying what called green policies. Those...

The question of energy savings has been a matter concern since long time in the mobile distributed systems and battery-constrained systems. However, for large-scale non-mobile systems, which nowadays reach impressive sizes, dimension (electrical consumption) just starts to be taken into account. In this paper, we present GREEN-NET <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> framework is based on 3 main components: an ON/OFF model...

Energy efficiency in large-scale distributed systems has recently emerged as a hot topic. This paper addresses some theoretical and experimental aspects of energy by putting perspective assumptions made this domain observations analyses. Based on results measurements, we revisit focus "truths" commonly assumed concerning the usage servers, links between resource load consumed energy, impact ON/OFF models, wrong linking virtualization.

Virtualized data centers where several virtual machines (VMs) are hosted per server becoming more popular due to Cloud Computing. As a consequence of energy efficiency concerns, the exact combination VMs running on specific will most likely change over time. We present experimental results how use energy/power consumption logs power monitored as side-channel that allows us recognize it currently hosts high degree. For classification, we maximum log-likelihood approach, which works well for...

Monitoring and assessing the energy efficiency of supercomputers data centers is crucial in order to limit reduce their consumption. Applications from domain High Performance Computing (HPC), such as MPI applications, account for a significant fraction overall consumed by HPC centers. Simulation popular approach studying behavior these applications variety scenarios, it therefore advantageous be able study consumption cost-efficient, controllable, also reproducible simulation environment....

This paper presents an integrated framework for energy savings in large scale distributed systems such as grids and clouds. The comprises tools mechanisms: to measure log data about the consumed by resources; present this information users; involve users decisions reduce their applications; enforce reduction automatically while respecting users' requirements achieving resource availability demanded current services. Experiments demonstrate achieved proposed mechanisms explore trade-offs...

The fast growth of cloud computing considerably increases the energy consumption infrastructures, especially, data centers. To reduce brown and carbon footprint, renewable such as solar/wind is considered recently to supply new green As intermittent fluctuates from time time, this paper considers two fundamental approaches for improving usage in a small/medium-sized center. One approach based on opportunistic scheduling: more jobs are performed when available. other relies Energy Storage...

The costs of current data centers are mostly driven by their energy consumption (specifically the air conditioning, computing and networking infrastructure). Yet, pricing models usually static rarely consider facilities' per user. challenge is to provide a fair predictable model attribute overall virtual machine (VM). Current pay-as-you-go Cloud providers allow users easily know how much will cost. However, this not fully transparent as where come from (e.g., energy). In paper we introduce...

Cloud computing allows for elasticity as users can dynamically benefit from new virtual resources when their workload increases. Such a feature requires highly reactive resource provisioning mechanisms. In this paper, we propose two prediction models, based on constraint programming and neural networks, that be used dynamic in environments. We also present trace generators help to extend an experimental dataset order test more widely optimization heuristics. Our models are validated using...

Wired networks are increasing in size and their power consumption is becoming a matter of concern. Evaluating the end-to-end electrical cost new network architectures protocols difficult due to lack monitored realistic infrastructures. We propose an End-to-End energy Cost mOdel simulator For large-scale Networks (ECOFEN) whose user's entries topology traffic. Based on configurable measurement different components (routers, switches, NICs, etc.), it provides overall including end-hosts as...

The fast growth of demand for computing and storage resources in data centers has considerably increased their energy consumption. Improving the utilization center integrating renewable energy, such as solar wind, been proposed to reduce both brown consumption carbon footprint centers. In this paper, we propose a novel framework oPportunistic schedulIng broKer infrAstructure (PIKA) save small mono-site order consumption, PIKA integrates resource overcommit techniques that help minimize...

Large-scale distributed systems (high-performance computing centers, networks, data centers) are expected to consume huge amounts of energy. In order address this issue, shutdown policies constitute an appealing approach able dynamically adapt the resource set actual workload. However, multiple constraints have be taken into account for such applied on real infrastructures: time and energy cost switching off, power consumption bounds caused by electricity grid or cooling system, availability...

The question of energy savings has been a matter concern since long time in the mobile distributed systems and battery-constrained systems. However, for large-scale non-mobile systems, which nowadays reach impressive sizes, dimension (electrical consumption) just starts to be taken into account. In this paper, we analyze usage an experimental grid over one-year period. Based on analysis, propose resource reservation infrastructure takes account issue. We validate our large scale Grid5000...