As the demand for AI-driven workloads increases, the energy consumption of Graphics Processing Units (GPUs) devices has come under intense scrutiny, particularly in hyperscale data centers where large numbers of accelerators are centralized and leased to diverse clients.

In an effort to raise awareness on the increasing carbon emissions of Cloud computing, the European Corporate Sustainability Reporting Directive effectively requires providers to supply their customers with an assessment of the carbon impact associated with their use. This represents a challenge for bare metal servers, where the deployment of dedicated power meters is often unfeasible at scale. To address this, we present PPEM-BM, a novel sensor-driven modeling approach to estimate the power consumption of bare metal servers using CPU temperature data acquired via IPMI. PPEM-BM enhances and generalizes the existing PowerHeat method, which correlates CPU temperature with power. Our methodology involves training individual power models, performing cross-evaluation to determine their portability, and then using a Learning to Rank (LTR) model to select the most appropriate pre-trained model for a target server based on its hardware configuration and CPU temperature statistics. An experiment conducted on$\mathbf{1, 0 7 6}$production servers at OVHcloud shows that PPEM-BM demonstrates a significant improvement compared to models based solely on hardware profiles. The approach offers a practical, scalable, and cost-effective solution for hosting providers to monitor energy consumption without widespread sensor deployment.

The Internet of Things (IoT) is used for various domains such as monitoring the environment, health care, and smart cities. Monitoring and measuring energy consumption of these systems is a crucial step in making them energy efficient. External Hardware-based power monitoring is not always available for IoT edge nodes. An alternative is to create an accurate power model that relates easy-to-monitor parameters (e.g., instructions count, cache misses, node temperature, etc) to externally monitored power. This relationship helps to estimate the power drawn by the nodes. IoT edge nodes have several power optimization leverages like Dynamic Voltage and Frequency Scaling (DVFS). When models calibration does not consider these leverages, the gap between power estimation and actual power usage increases. In related works, several power models and corresponding Software-defined power meters do not consider CPU frequency on IoT edge nodes. These Software-defined power meters provide regression-based power models for IoT edge nodes. This work compares predictions made by these state of the art power models to accurate external power monitoring. We show that not considering CPU frequency can result in incorrect estimations. We investigate and compare several methodologies for building power models, considering the CPU frequency, power, and energy leverage. Different performance metrics and regression methods are explored to estimate power usage. We demonstrate that linear and polynomial regression-based models are able to account for various CPU frequencies on IoT edge nodes. Using these models, we can predict the power consumed by IoT edge nodes running a specific workload, with a MAPE of 2% compared to accurate Hardware-based power meters.

Information and Communication Technologies (ICT) and Data Centres (DC) have nowadays considerable environmental impacts. The number of applications and services hosted in the cloud is significant, and the associated number of infrastructure is steadily increasing. Cloud Service Providers (CSP) need to respond to this growing demand and size their infrastructures accordingly but users misbehaviour and expectations in terms of service quality lead to oversized infrastructures. Infrastructures are sized to meet peak-demand, resulting in poor resource utilization and additional power consumption. But as their behaviour can increase DC energy consumption and environmental footprint, users can also help reducing them. In this paper, we study how users provided with a simple tool can reduce the power consumption of a virtualization cluster in a cloud company. Using a Virtual Machine (VM) shutdown policy, users can directly contribute to the mitigation of the power consumption of the infrastructure. Part of the paper is dedicated to profile the users and understand their behaviour when it comes to powering off their VMs. To further reduce the energy consumption of the cluster, we combine the VM shutdown policy with a simple consolidation heuristic. Simulations show a 23.95% power consumption reduction, with an additional 8.72% reduction thanks to the users. A production implementation was conducted and results in a 12.58% power consumption reduction over one week.

EcoIndex has been proposed to evaluate the absolute environmental performance of a given URL using a score ranging from 0 to 100 (the higher, the better). In this article, we make a critical analysis of the initial approach and propose alternatives that no longer calculate a plain score but allow the query to be situated among other queries. The generalized critiques come with statistics and rely on extensive experiments (first contribution). Then, we move on to low-cost Machine Learning (ML) approaches (second contribution) and a transition before obtaining our final results (third contribution). Our research aims to extend the initial idea of analytical computation, i.e., a relation between three variables, in the direction of algorithmic ML computations. The fourth contribution corresponds to a discussion on our implementation, available on a GitHub repository. Along with the paper, we invite the reader to examine the question: What attributes make sense for our problem?, or equivalently, what is a relevant data policy for studying digital environmental impacts? Beyond computational questions, it is important for the scientific community to focus on this question in particular. We currently promote using well-established ML techniques because of their potential, which we discuss in the paper. However, we also question techniques for their frugality or otherwise. Our data science project is still at the data exploration stage. We also want to encourage synergy between technical expertise and business knowledge because this is fundamental for advancing the data project.

The global energy demand for digital activities is constantly growing. Computing nodes and cloud services are at the heart of these activities. Understanding their energy consumption is an important step towards reducing it. On one hand, physical power meters are very accurate in measuring energy but they are expensive, difficult to deploy on a large scale, and are not able to provide measurements at the service level. On the other hand, power models and vendor-specific internal interfaces are already available or can be implemented on existing systems. Plenty of tools, called software-based power meters, have been developed around the concepts of power models and internal interfaces, in order to report the power consumption at levels ranging from the whole computing node to applications and services. However, we have found that it can be difficult to choose the right tool for a specific need. In this work, we qualitatively and experimentally compare several software-based power meters able to deal with CPU or GPU-based infrastructures. For this purpose, we evaluate them against high-precision physical power meters while executing various intensive workloads. We extend this empirical study to highlight the strengths and limitations of each software-based power meter.

Internet of Things (IoT) applications often require the processing of data streams generated by devices dispersed over a large geographical area. Traditionally, these data streams are forwarded to a distant cloud for processing, thus resulting in high application end-to-end latency. Recent work explores the combination of resources located in clouds and at the edges of the Internet, called cloud-edge infrastructure, for deploying Data Stream Processing (DSP) applications. Most previous work, however, fails to scale to very large IoT settings. This paper introduces deployment strategies for the placement of Data Stream Processing (DSP) applications onto cloud-edge infrastructure. The strategies split an application graph into regions and consider regions with stringent time requirements for edge placement. The proposed Aggregate End-to-End Latency Strategy with Region Patterns and Latency Awareness (AELS+RP+LA) decreases the number of evaluated resources when computing an operator's placement by considering the communication overhead across computing resources. Simulation results show that, unlike the state-of-the-art, Aggregate End-to-End Latency Strategy with Region Patterns and Latency Awareness (AELS+RP+LA) scales to environments with more than 100k resources with negligible impact on the application end-to-end latency.

This paper compares the cooling-efficiency between air and immersion cooling methods through the implementation of an experimental framework. Short and long term stresses with and without overclocking were run and monitored on 8 identical HPC servers. A statistical analysis using the Mann-Whitney method was conducted to determine whether or not, the temperature differences observed between the servers cooled by immersion and the ones cooled by air were significant. The results were promising with immersion cooling achieving better overall cooling performance with low energy consumption as no cooling units were used for the immersion cooling system. A discussion on practical issues encountered using the immersion technology is also included.

Despite recent advances in improving the performance of high performance computing (HPC) and distributed systems, power dissipation and thermal cooling challenges persist, impacting their total cost of ownership. Making HPC systems more energy and thermal efficient will require understanding of individual power dissipation and temperature contributions of multiple hardware system components and their accompanying software. In this work, we present an experimental workflow for energy and temperature profiling on systems running parallel applications. It allows full and dynamic control over the execution of applications for the entire frequency range. Through its use, we show that the energy response to frequency scaling is highly dependent on the workload characteristics and it is convex in nature with an optimal frequency point. During the course of our experimentation, we encountered a non-intuitive finding, where we observed that the tested low-power processor is consuming more power on average than the standard processor.

Building a sustainable Exascale machine is a very promising target in High Performance Computing (HPC). To tackle the energy consumption challenge while continuing to provide tremendous performance, the HPC community have rapidly adopted GPU-based systems. Today, GPUs have became the most prevailing components in the massively parallel HPC landscape thanks to their high computational power and energy efficiency. Modeling the energy consumption of applications running on GPUs has gained a lot of attention for the last years. Alas, the HPC community lacks simple yet accurate simulators to predict the energy consumption of general purpose GPU applications. In this work, we address the prediction of the energy consumption of CUDA kernels via simulation. We propose in this paper a simple and lightweight energy model that we implemented using the open-source framework SimGrid. Our proposed model is validated across a diverse set of CUDA kernels and on two different NVIDIA GPUs (Tesla M2075 and Kepler K20Xm). As our modeling approach is not based on performance counters or detailed-architecture parameters, we believe that our model can be easily approved by users who take care of the energy consumption of their GPGPU applications.

The number of Internet of Things applications is forecast to grow exponentially within the coming decade. Owners of such applications strive to make predictions from large streams of complex input in near real time. Cloud-based architectures often centralize storage and processing, generating high data movement overheads that penalize real-time applications. Edge and Cloud architecture pushes computation closer to where the data is generated, reducing the cost of data movements and improving the application response time. The heterogeneity among the edge devices and cloud servers introduces an important challenge for deciding how to split and orchestrate the IoT applications across the edge and the cloud. In this paper, we extend our IoT Edge Framework, called R-Pulsar, to propose a solution on how to split IoT applications dynamically across the edge and the cloud, allowing us to improve performance metrics such as end-to-end latency (response time), bandwidth consumption, and edge-to-cloud and cloud-to-edge messaging cost. Our approach consists of a programming model and real-world implementation of an IoT application. The results show that our approach can minimize the end-to-end latency by at least 38% by pushing part of the IoT application to the edge. Meanwhile, the edge-to-cloud data transfers are reduced by at least 38% and the messaging costs are reduced by at least 50% when using the existing commercial edge cloud cost models.

There is increasing demand for handling massive amounts of data in a timely manner via Distributed Stream Processing (DSP). A DSP application is often structured as a directed graph whose vertices are operators that perform transformations over the incoming data and edges representing the data streams between operators. DSP applications are traditionally deployed on the Cloud in order to explore the virtually unlimited number of resources. Edge computing has emerged as a suitable paradigm for executing parts of DSP applications by offloading certain operators from the Cloud and placing them close to where the data is generated, hence minimising the overall time required to process data events (i.e., the end-to-end latency). The operator reconfiguration consists of changing the initial placement by reassigning operators to different devices given target performance metrics. In this work, we model the operator reconfiguration as a Reinforcement Learning (RL) problem and define a multi-objective reward considering metrics regarding operator reconfiguration, and infrastructure and application improvement. Experimental results show that reconfiguration algorithms that minimise only end-to-end processing latency can have a substantial impact on WAN traffic and communication cost. The results also demonstrate that when reconfiguring operators, RL algorithms improve by over 50% the performance of the initial placement provided by state-of-the-art approaches.

Distributed Stream Processing (DSP) applications are increasingly used in new pervasive services that process enormous amounts of data in a seamless and near real-time fashion. Edge computing has emerged as a means to minimise the time to handle events by enabling processing (i.e., operators) to be offloaded from the Cloud to the edges of the Internet, where the data is often generated. Deciding where to execute such operations (i.e., edge or cloud) during application deployment or at runtime is not a trivial problem. In this work, we employ Reinforcement Learning (RL) and Monte-Carlo Tree Search (MCTS) to reassign operators during application runtime. Experimental results show that RL and MCTS algorithms perform better than traditional placement techniques. We also introduce an optimisation to a MCTS algorithm, called MCTS-Best-UCT, that achieves similar latency with fewer operator migrations and faster execution time. In certain scenarios, the time needed by MCTS-Best-UCT to find the best end-to-end latency is at least 33% smaller than the time required by the other algorithms.

Combining computation and water warming is an efficient alternative for recycling dissipated energy. Based on a set of Computing Water Heaters, used as a distributed super computer, we introduce a computing water heater scheduling system that has been validated by simulation on extreme workloads. These simulations lead us to reach an approximate CPU usage of about 60% while minimizing the thermal resistance usage to 10% (only when necessary) on a complex steady state.

Summary Current large‐scale systems, like datacenters and supercomputers, are facing an increasing electricity consumption. These infrastructures are often dimensioned according to the workload peak. However, as their consumption is not power‐proportional when the workload is low, the power consumption is still high. Shutdown techniques have been developed to adapt the number of switched‐on servers to the actual workload. However, datacenter operators are reluctant to adopt such approaches because of their potential impact on reactivity and hardware failures, and their energy gain is often largely misjudged. In this article, we evaluate the potential gain of shutdown techniques by taking into account shutdown and boot up costs in time and energy. This evaluation is made on recent server architectures and future energy‐aware architectures. Our simulations exploit real traces collected on production infrastructures under various machine configurations with several shutdown policies, with and without workload prediction. We study the impact of future's knowledge for saving energy with such policies. Finally, we examine the energy benefits brought by suspend‐to‐disk and suspend‐to‐RAM techniques, and we study the impact of shutdown techniques on the energy consumption of prospective hardware with heterogeneous processors (big‐medium‐little paradigm).

Energy consumption has become one of the mostcritical issues in the evolution of High Performance Computingsystems (HPC). Controlling the energy consumption of HPCplatforms is not only a way to control the cost but also a stepforward on the road towards exaflops. Powercapping is a widelystudied technique that guarantees that the platform will notexceed a certain power threshold instantaneously but it givesno flexibility to adapt job scheduling to a longer term energybudget control. We propose a job scheduling mechanism that extends thebackfilling algorithm to become energy-aware. Simultaneously, we adapt resource management with a node shutdown technique to minimize energy consumption whenever needed. Thiscombination enables an efficient energy consumption budgetcontrol on a cluster during a period of time. The technique isexperimented, validated and compared with various alternativesthrough extensive simulations. Experimentation results show highsystem utilization and limited bounded slowdown along withinteresting outcomes in energy efficiency while respecting anenergy budget during a particular time period.

Traffic engineering technologies such as MPLS have been proposed to adjust the paths of data flows according to network availability. Although the time interval between traffic optimisations is often on the scale of hours or minutes, modern SDN techniques enable reconfiguring the network more frequently. It is argued, however, that changing the paths of TCP flows too often could severely impact their performance by incurring packet loss and reordering. This work analyses and evaluates the impact of frequent route changes on the performance of TCP flows. Experiments carried out on a network testbed show that rerouting a flow can affect its throughput when reassigning it a path either longer or shorter than the original path. Packet reordering has a negligible impact when compared to the increase of RTT. Moreover, constant rerouting influences the performance of the congestion control algorithm. Designed to assess the limits on SDN-induced reconfiguration, a scenario where the traffic is rerouted every 0.1s demonstrates that the throughput can be as low as 35% of that achieved without rerouting.

Networking infrastructures are considered to consume as much energy as terminal end-user equipment or datacenters. While energy consumption of wireless networks is a matter of concern since their beginning, it is not the case for wired networks as they do not rely on batteries, but on plugged equipment. Yet, facing growing consumption, energy-efficient techniques start to be implemented in wired networks. However, measuring the end-to-end energy consumption of wired networking infrastructures remains a real challenge for network operators and scientists. This article presents the ECOFEN (Energy Consumption mOdel For End-to-end Networks) framework which allows to support precise simulation of energy consumption of large-scale complex wired networks. The experimental validation shows that Ecofen provides accurate energy consumption values.

Summary The adoption of cloud computing is still limited by several legal concerns that customers may have, such as data sovereignty. In cloud computing, data can be physically hosted in sensible locations, resulting in a lack of control for companies. In this context, we present the Nu@ge project, which aims at building a federation of container‐sized datacenters in the French territory. Nu@ge provides a software stack that enables companies to interconnect independent datacenters in a national mesh. A software architecture is presented and implemented as a federation of small datacenters deployed in France. The proposed architecture enables cooperation between local customized‐cloud managers and a federation‐wide middleware. It uses monitoring information from facilities and performance indicators from physical servers for managing the system, preventing incidents and considering energy efficiency. Additionally, a prototype of a container‐sized datacenter has been validated and patented.

The recent rapid expansion of Cloud computing facilities triggers an attendant challenge to facility providers and users for methods for optimal placement of workflows on distributed resources, under the often-contradictory impulses of minimizing makespan, energy consumption, and other metrics. Evolutionary Optimization techniques that from theoretical principles are guaranteed to provide globally optimum solutions, are among the most powerful tools to achieve such optimal placements. Multi-Objective Evolutionary algorithms by design work upon contradictory objectives, gradually evolving across generations towards a converged Pareto front representing optimal decision variables — in this case the mapping of tasks to resources on clusters. However the computation time taken by such algorithms for convergence makes them prohibitive for real time placements because of the adverse impact on makespan. This work describes parallelization, on the same cluster, of a Multi-Objective Differential Evolution method (NSDE-II) for optimization of workflow placement, and the attendant speedups that bring the implicit accuracy of the method into the realm of practical utility. Experimental validation is performed on a reallife testbed using diverse Cloud traces. The solutions under different scheduling policies demonstrate significant reduction in energy consumption with some improvement in makespan.

As the number of data centers increases, it is urgent to reduce their energy consumption. Although servers are becoming more energy-efficient, their idle consumption remains high, which is an issue as data centers are often over-provisioned. This work proposes a novel approach for building data centers with heterogeneous machines carefully chosen for their performance and energy efficiency ratios. We focus on web applications whose load varies over time, and design a scheduler that dynamically reconfigures the infrastructure, by migrating applications and switching machines on or off, so that the energy consumed by the data center is proportional to the load. Experiments evaluate the approach and show the energy savings achieved by our heterogeneous data center design and management while satisfying Quality of Service (QoS) constraints.

This work investigates factors that can impact the elasticity of bare-metal resources. We analyse data from a real bare-metal deployment system to build a deployment time model, which is used to evaluate how provisioning time impacts the reservation of bare-metal resources. Climate/Blazar, a reservation framework designed for OpenStack, is discussed. Simulation results show that reservations can help reduce the time to deliver a provisioned cluster to its customer while achieving energy savings similar to those of strategies that switch-off idle resources.

In distributed model predictive control, accounting for the interactions of subsystems requires exchanging information between controllers. However, challenges arise when controllers may lose the synchronization and when shared information may encounter the divergence due to the switching topology and the imperfection in communication. Managing divergent (or outdated) information is considered in this paper as a consensus problem and solved by an asynchronous consensus protocol. This approach based on multi-agent system paradigm to distributed control requires each controller to agree on some data values needed during action computation with its neighbors. Under certain assumptions, the convergence of consensus protocol is analytically obtained with various communication patterns (for example, directed and delayed communication). As a result, the performance of control system is improved in comparison with the decentralized strategy, and approached to the distributed strategy with sequential timing. Some simulation examples are given to illustrate the efficiency of this approach for the control of irrigation canals modeled by the Lattice Boltzmann method.

Reducing the energy consumed by wired computer networks is a challenge that has been actively investigated over the past few years. A popular mechanism proposed to reduce the consumption aims to put links and line cards to sleep mode during off-peak hours. Such a mechanism, however, decreases the available network capacity and increases the risk of congestion if traffic rises unexpectedly. This paper proposes a solution to rapidly react to network bursts and turn-on sleeping links, which we term as SegmenT Routing based Energy Efficient Traffic Engineering for switching ON (STREETE-ON). The proposed algorithm was implemented in the OMNeT++ network simulator using state-of-art dynamic graph algorithms. In such a way, we achieved execution times of tens of milliseconds for a 50-node network. Experimental results show that STREETE-ON can effectively prevent network congestion, avoid turning-on unneeded links, and preserve good energy-efficiency of the network.

The increasing number of data centers raises serious concerns regarding their energy consumption. Although servers have become more energy-efficient over time, their idle consumption remains high, which is an issue as resources in data centers are often over-provisioned. This work proposes a novel approach for building data centers so that their energy consumption is proportional to load. A data center hence comprises heterogeneous machines carefully chosen for their performance and energy efficiency ratios. We focus on web applications whose load varies over time and design a scheduler that dynamically reconfigures the infrastructure to minimize its energy consumption according to current load and application requirements. Based on load forecasts, it takes reconfiguration decisions and performs actions such as migrating applications and switching machines on or off. The approach is evaluated considering a data center with heterogeneous resources, and the experiments show how to adjust the parameters of scheduling policies to save the most energy while satisfying Quality of Service (QoS) constraints.

In a nutshell, this paper presents a dynamic routing design for control applications using a hybrid network architecture. For application-specific routing requirements, we propose a weighted composition of few primary metrics in such a way that the routing protocol using the resulting composite metric achieves convergence, loop-free and path-optimal properties. From this composition method, we point out the relation between combined routing metrics and network performance criteria. Thereafter, the composite metric can be adapted in order to satisfy requirements of control systems in terms of communication. Our approach is then demonstrated by applying composite metrics with the Routing Protocol for Low-power and lossy network (RPL), specified by Internet Engineering Task Force (IETF). By combining some primary metrics quantified in their standard definitions, different network performance criteria are evaluated and compared through extensive computer simulation scenarios.

Energy consumption is one of the main limiting factors for the development and deployment of large scale distributed systems (datacenters, Clouds, networks). A lot of academic and industrial researches propose several hardware and software based approaches in order to reduce energy consumption and to exhibit energy proportional and energy efficient capabilities. But for reaching such improvements a full vision of environmental impact of ICT must be taken into account: from material extraction, transport to design, usage and end of life. This talk will explore some challenges in dealing with ICT life cycle and will illustrate that with several scenario and examples.

The huge amount of energy consumed by data centers represents a limiting factor in their operation. Many of these infrastructures are over-provisioned, thus a significant portion of this energy is consumed by inactive servers staying powered on even if the load is low. Although servers have become more energy-efficient over time, their idle power consumption remains still high. To tackle this issue, we consider a data center with an heterogeneous infrastructure composed of different machine types - from low power processors to classical powerful servers - in order to enhance its energy proportionality. We develop a dynamic provisioning algorithm which takes into account the various characteristics of the architectures composing the infrastructure: their performance, energy consumption and on/off reactivity. Based on future load information, it makes intelligent decisions of resource reconfiguration that impact the infrastructure at multiple terms. Our algorithm is reactive to load evolutions and is able to respect a perfect Quality of Service (QoS) while being energy-efficient. We evaluate our original approach with profiling data from real hardware and the experiments show that our dynamic provisioning brings significant energy savings compared to classical data centers operation.

These workshops were organized by members of the Nesus Cost Action IC 1305: Network for Sustainable Ultrascale Computing, which is a follow-up of COST Actions IC0804 and IC0805 1. The goal of the NESUS Action is to establish an open European research network targeting sustainable solutions for ultrascale computing aiming at cross fertilization among HPC, large-scale distributed systems, and big data management. This network aims at contributing to glue disparate researchers working across different areas and provide a meeting ground for researchers in these separate areas to exchange ideas, to identify synergies, and to pursue common activities in research topics such as sustainable software solutions (applications and system software stack), data management, energy efficiency, and resilience. The selected papers cover very important scientific issues encountered nowadays such as the following: CPU/GPU execution, system-on-chip programming, parallel algorithms taking into account various constraints (energy, communication, etc.), programming models, and so on. We really hope that the reader will enjoy this high-quality issue, and we are sure that she/he will find it highly relevant to the state-of-the-art of today's heterogeneous and parallel computing.

The subsystems that compose a HPC platform (e.g. CPU, memory, storage and network) are often designed and configured to deliver exceptional performance to a wide range of workloads. As a result, a large part of the power that these subsystems consume is dissipated as heat even when executing workloads that do not require maximum performance. Attempts to tackle this problem include technologies whereby operating systems and applications can reconfigure subsystems dynamically, such as by using DVFS for CPUs, LPI for network components, and variable disk spinning for HDDs. Most previous work has explored these technologies individually to optimise workload execution and reduce energy consumption. We propose a framework that performs on-line analysis of an HPC system in order to identify application execution patterns without a priori information of their workload. The framework takes advantage of reoccurring patterns to reconfigure multiple subsystems dynamically and reduce overall energy consumption. Performance evaluation was carried out on Grid'5000 considering both traditional HPC benchmarks and real-life applications.

This paper focuses on energy optimisation in the context of lifts. Modern lifts embed batteries that are so far used only in emergency. We propose a multi-level optimisation strategy to reduce the electricity bill by combining harvesting, the grid and energy stored in batteries. The strategy combines several analytic components (forecaster, optimisers), modelled/measured variables, and is used by the control system. A coordination middleware enables the cooperation between the components embedded in the lift or located in the cloud, thus requiring communication through firewalls of different companies. Early results are presented. They illustrate new features for improving energy efficiency and they demonstrate our capacity to build such an optimisation architecture in a real environment. Part of the results are simulated to extrapolate the reachable energy gain.

Joining the European Arrowhead Project, which looks forward to a modern collaborative automation, we will present in this positioning paper the irrigation network control problems and corresponding multi-layer approach. The first layer is the hydraulic network, itself represented by a complex model obtained in coupling Shallow Water Equations for free surface flows and Lattice Boltzmann Method to get a tractable model for optimization and supervision purposes. The second layer is the heterogeneous communication network using hybrid architectures and 6LoWPAN, a unified protocol for wired and wireless sensor networks. The third layer is the optimal reactive control system, itself developed using methods from decentralized artificial intelligent systems (namely multi-agent systems). Detailed discussions of each layer with some analytical results will be described in this paper. We will outline the potential interest of the multi-layer approach, more precisely its efficiency and reliability for supervision, energy optimization and hydraulic control of complex water transportation systems.

In this paper are presented some preliminary results on an autonomous lift subsystem made with the interconnection of a DC/AC controlled converter, a salient permanent magnet synchronous motor (SPMSM), the charge and counterweight and the mechanical transmission system. These results are firstly the derivation of a Bond Graph and a related Port-Controlled Hamiltonian (PCH) models and, secondly, the analysis of two identification methods to estimate online the SPMSM parameters.

SUMMARY Nowadays, there is no doubt that energy consumption has become a limiting factor in the design and operation of high performance computing (HPC) systems. This is evidenced by the rise of efforts both from the academia and the industry to reduce the energy consumption of those systems. Unlike hardware solutions, software initiatives targeting HPC systems' energy consumption reduction despite their effectiveness are often limited for reasons including (1) the program specific nature of the solution proposed; (2) the need of deep understanding of the application task at hand; and (3) proposed solutions are often difficult to use by novices and/or are designed for single task environments. This paper proposes a three step blind system‐wide, application independent, fine‐grain, and easy to use (user friendly) methodology for improving energy performance of HPC systems. The methodology typically breaks into phase detection, phase characterization, and phase identification and system reconfiguration. And it is blind in the sense that it does not require any knowledge from users. It relies upon reconfigurable capabilities offered by the majority of HPC subsystems—including the processor, storage, memory, and communication subsystems—to reduce the overall energy consumption of the system (excluding network equipments) at runtime. We also present an implementation of our methodology through which we demonstrate its effectiveness via static analyses and experiments using benchmarks representative of HPC workloads. The memory becoming one of the most power hungry HPC system this paper also introduces and investigates the relevance of a potential power saving scheme which we refer to as memory size scaling. It is destined to scale the memory size for saving energy in a CPU frequency scaling like fashion. Copyright © 2014 John Wiley & Sons, Ltd.

Energy consumption and fault tolerance are two interrelated issues to address for designing future exascale systems. Fault tolerance protocols used for check pointing have different energy consumption depending on parameters like application features, number of processes in the execution and platform characteristics. Currently, the only way to select a protocol for a given execution is to run the application and monitor the energy consumption of different fault tolerance protocols. This is needed for any variation of the execution setting. To avoid this time and energy consuming process, we propose an energy estimation framework. It relies on an energy calibration of the considered platform and a user description of the execution setting. We evaluate the accuracy of our estimations with real applications running on a real platform with energy consumption monitoring. Results show that our estimations are highly accurate and allow selecting the best fault tolerant protocol without pre-executing the application.

Checkpointing protocols have different energy consumption depending on parameters like application features and platform characteristics. To select a protocol for a given execution, we propose an energy estimator that relies on an energy calibration of the considered platform and a user description of the execution settings.

Future supercomputers will gather hundreds of millions of communicating cores. The movement of data in such systems will be very energy consuming. We address in this paper the issue of energy consumption of data broadcasting in such large scale systems. To this end, we propose a framework to estimate the energy consumed by different MPI broadcasting algorithms for various execution settings. Validation results show that our estimations are highly accurate and allow to select the least consuming broadcasting algorithm.

The paper describes different research concepts of which the combination leads to an improvement of the power consumption per subscriber in a wireline access network by 10× and the energy efficiency per transferred bit by 100×.

The rising computing demands of scientific endeavors often require the creation and management of High Performance Computing (HPC) systems for running experiments and processing vast amounts of data. These HPC systems generally operate at peak performance, consuming a large quantity of electricity, even though their workload varies over time. Understanding the behavioral patterns (i.e., phases) of HPC systems during their use is key to adjust performance to resource demand and hence improve the energy efficiency. In this paper, we describe (i) a method to detect phases of an HPC system based on its workload, and (ii) a partial phase recognition technique that works cooperatively with on-the-fly dynamic management. We implement a prototype that guides the use of energy saving capabilities to demonstrate the benefits of our approach. Experimental results reveal the effectiveness of the phase detection method under real-life workload and benchmarks. A comparison with baseline unmanaged execution shows that the partial phase recognition technique saves up to 15% of energy with less than 1% performance degradation.

Modern high performance computing subsystems (HPC) – including processor, network, memory, and IO – are provided with power management mechanisms. These include dynamic speed scaling and dynamic resource sleeping. Understanding the behavioral patterns of high performance computing systems at runtime can lead to a multitude of optimization opportunities including controlling and limiting their energy usage. In this paper, we present a general purpose methodology for optimizing energy performance of HPC systems considering processor, disk and network. We rely on the concept of execution vector along with a partial phase recognition technique for on-the-fly dynamic management without any a priori knowledge of the workload. We demonstrate the effectiveness of our management policy under two real-life workloads. Experimental results show that our management policy in comparison with baseline unmanaged execution saves up to 24% of energy with less than 4% performance overhead for our real-life workloads.

Virtualized data centers where several virtual machines (VMs) are hosted per server are becoming more popular due to Cloud Computing. As a consequence of energy efficiency concerns, the exact combination of VMs running on a specific server will most likely change over time. We present experimental results how to use the energy/power consumption logs of a power monitored server as a side-channel that allows us to recognize the exact combination of VMs it currently hosts to a high degree. For classification, we use a maximum log-likelihood approach, which works well for comparably small training and test set sizes. We also show to which degree a specific VM can be recognized, regardless of other VMs currently running on the same server, and show false negative/positive rates. To cross-validate our results, we have used a Kolmogorov-Smirnov test, resulting in comparable quality of recognition within shorter time. In order to clarify whether our approach is generalizable and yields reproducible results, we have set up a second experimental infrastructure in Lyon, using a different hardware platform and power measurement device. We have obtained similar results and have experimented with different CPU frequency scaling governors, yielding comparable quality of recognition. As a result, energy consumption data of servers must be protected carefully, as it is potentially valuable information for an attacker trying to track down a VM to mount further attack steps.

In this short paper, we introduce the new French Cloud management software called Compatible One. The intent of this project is to build an innovative Cloud framework based on well known standards, existing software and technologies. Given our approach it allows us to exploit original solutions and moreover include at design level some new paradigms like energy efficiency. This paper focuses on the current activities done in the Compatible One project for energy monitoring and energy efficient management of Clouds systems.

Reducing the energy consumption of wired networks has become a key concern for manufacturers of network equipments and network providers. In this paper, we propose an energy-efficient data transfer framework that uses advance bandwidth provisioning and on/off algorithms to put unused nodes of wired networks into sleep mode. This framework is termed as High-level Energy-awaRe Model for bandwidth reservation in End-to-end networkS (HERMES). We explore centralized, decentralized and clustered approaches for managing network resources and bandwidth reservations in this framework. By applying these approaches to HERMES, we evaluate via simulation their efficiency in terms of performance and energy consumption.

This paper presents HERMES: an energy-efficient data transfer framework for data-center, grid and cloud networks. An architecture and simulations are provided to show that this framework could save more than two-thirds of the energy currently consumed by these networks.

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

The current Internet does not enable easy introduction and deployment of new network technologies and services. This paper aims to progress the Future Internet (FI) by introduction of a service composition and execution environment that re-use existing components of access and core networks. This paper presents essential service-centric platforms and software systems that have been developed with the aim to create a flexible environment for an Autonomic Internet.

This article presents the design and building of DSL-Lab, a platform to experiment on distributed computing over broadband domestic Internet. Experimental platforms such as PlanetLab and Grid'5000 are promising methodological approaches to study distributed systems. However, both platforms focus on high-end service and network deployments only available on a restricted part of the Internet, leaving aside the possibility for researchers to experiment in conditions close to what is usually available with domestic connection to the Internet. DSL-Lab is a complementary approach to PlanetLab and Grid'5000 to experiment with distributed computing in an environment closer to how Internet appears, when applications are run on end-user PCs. DSL-Lab is a set of 40 low-power and low-noise nodes, which are hosted by participants, using the participants' xDSL or cable access to the Internet. The objective is to provide a validation and experimentation platform for new protocols, services, simulators and emulators for these systems. In this paper, we report on the software design (security, resources allocation, power management) as well as on the first experiments achieved.

International audience

The question of energy savings has been a matter of concern since a long time in the mobile distributed systems and battery-constrained systems. However, for large-scale non-mobile distributed systems, which nowadays reach impressive sizes, the energy dimension (electrical consumption) just starts to be taken into account. In this paper, we present the GREEN-NET1framework which is based on 3 main components: an ON/OFF model based on an Energy Aware Resource Infrastructure (EARI), an adapted Resource Management System (OAR) for energy efficiency and a trust delegation component to assume network presence of sleeping nodes.

For an improved QoS provisioning, operators are strongly concerned both with the availability and ther eliability of their performance critical servers. Among these entities, we mention the firewalls used to filter the offered network traffic to the operator's information system as well as the end processing servers which service each incoming client request. In this work, we advocate and evaluate a full architecture for highly available services. Performance evaluations show that our proposed framework incurs a minimal overhead to the end-to-end communications during failsafe periods and performs well during failures.

Delivering live streaming content over the Internet requires low delay and smooth packet transmission rate. TCP introduces rate oscillations and requires more buffering and bandwidth to sustain uninterrupted playback. In this paper we propose a new framework which facilitates streaming flows. Our solution provides a smoother rate control than TCP and improves streaming performance based on cross layer feedback between the transport protocol and streaming server. We present our experimental results through simulation.

Under overload condition, a processing server contributes to the poor QoS through sustaining heavy request queuing delays and prolonged processing time. Admission control is a well known mean to prevent a server overload. Most state of the art research advocate session oblivious mechanisms where the dropping of the requests pertaining to an accepted session can occur at any time during the session lifespan. From an operator perspective, this means that the server, which seems to sustain a high throughput, is in reality wasting its resources on failed or on reduced QoS sessions. In this work, we advocate an innovative architecture for session aware admission control of an offered IP traffic to a cluster-based server. The proposed system is enhanced with means for maximizing the useful throughput in terms of completed sessions per unit of time. It particularly achieves an improved responsiveness and a better stability. Finally, it is open to adapt to any multiple-flow based NGN service such as voice over IP or streaming video.

Failures are a permanent menace for the availability of Internet services. During the last decades, numerous fault-tolerant approaches have been proposed for the wide spectrum of Internet services, including stateful firewalls. Most of these solutions adopt reactive approaches to mask failures by replicating state-changes between replicas. However, reactive replication is a resource consuming task that reduces scalability and performance: the amount of computational and bandwidth resources to propagate state-changes among replicas might be high. On the other hand, more and more commercial off-the-shelf platforms provide integrated hardware error-detection facilities. As a result, some current fault-tolerance research works aim to replace the reactive fault-handling with proactive fault-avoidance. However, pure proactive approaches are risky and they currently face serious limitations. In this work, we propose a hybrid proactive and reactive model that exploits the stateful firewall semantics to increase the overall performance of cluster-based fault-tolerant stateful firewalls. The proposed solution reduces the amount of resources involved in the reactive state-replication by means of bayesian techniques to perform lazy replication while, at the same time, benefits from proactive fault-tolerance. Preliminary experimental results are also provided.

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

Fault tolerance can be defined as the capability of a system or a component to continue normal operation, despite the occurrence of a hardware or a software fault. Frameworks providing a fault tolerant service take advantage of resources' redundancy to provide high availability capabilities. One of the critical challenges that led to this work is the observation that the existing fault tolerance frameworks are not adapted to current and next generation Internet services. Indeed, they do not provide consistent service-aware failure recovery capabilities. Particularly, little interest has been granted to transport level awareness despite its important partaking in improving the reliability of connection oriented services and stateful devices. In this paper, we evaluate an active replication based framework for highly available Internet services. The proposed framework is fully client/server transparent. Performance evaluations show that it incurs minimal overhead to end-to-end conversations during failsafe periods and performs well during failures.

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

Stateful firewalls are security solutions widely deployed in the Internet. These devices filter network traffic and keep track of the state of connections in order to make the deployment of several attacks, such as TCP resets, difficult. However, firewalls are critical equipments in the network schema since they introduce a single point of failure. Therefore, a failure may isolate networks, users and interrupt established connections. Current fault tolerant solutions mask failures by means of replication techniques based on physical redundancy and state propagation. However, these solutions do not suit well for stateful firewall scenarios since they reduce bandwidth throughput roughly, they require costful extra hardware or are stuck to wasteful and inflexible single primary-backup settings. In this work we detail FT-FW (fault tolerant firewall), a software-based transparent connection failover mechanism for stateful firewalls. Our solution has a negligible impact in terms of performance, as well as the fact that quick recovery from failures and fast responses to clients are guaranteed. The architecture is suitable for low cost off-the-shelf systems and no extra hardware is required.

While a lot of researches focused on how to efficiently spread the offered network load on the available cluster resources, less interest has been granted to the impact of the used mechanisms on the reliable execution of the upper layer services. On the other hand, emerging NGN services as well as some of the already familiar services involve multiple flows during the lifespan of a single end- to-end session, hence, raising the challenge of session awareness while processing the incoming network traffic. In this paper, we grasp the need for fine grained session awareness to efficiently allocate the cluster resources to the offered network traffic. The analysis of load balancing scenarios of some representative IP services provides us with solid reasons to use deep packet inspection to achieve fine grained network traffic load distribution, and to meet NAT and firewall traversal constraints as well.

Today's Internet models are concerned with both of the quality and the availability of a given service. Current fault tolerant frameworks need to support consistent transport and application level failover mechanisms. Transport protocols relay on an explicit association between a service and its physical location for the wired Internet. Thus, when a host fails, the end-to-end flow terminates. In this paper, we propose an active replication based system which enhances the reliability of the already established TCP flows. The proposed scheme is client transparent. It does not incur any overhead to the end-to- end communication during failsafe periods, and performs well during failures.

The availability of some critical equipment like gateways, firewalls and proxies must be guaranteed in operational networks. In early equipments, the routing and filtering decisions were based on the packet information, nowadays this static approach is not longer safe. Existing high availability (HA) solutions do not cover all the aspects to ensure availability of advanced settings that are being deployed these days. Some important issues like the reduction of the downtime and the need for failure detection in such scenarios must be studied. This paper describes the implementation of high available stateful network equipments: these systems apply policies based on the state of the connections, such information is gathered in runtime by means of packet inspection. This work specifically focuses on Linux systems and firewalls because the IT industry trusts more and more OpenSource solutions to deploy critical services because of its quality and the access to the source code. We propose the SNE library (stateful network equipment), which is an add-on to current HA protocols, to solve the existing limitations. In this paper, we describe the proposed architecture and we detail a set problematic scenarios supported by our library, as well as first experiments and the evaluation.

XCP is a transport protocol that uses the assistance of specialized routers to very accurately determine the available bandwidth along the path from the source to the destination. In this way, XCP efficiently controls the sender's congestion window size thus avoiding the traditional slow-start and congestion avoidance phase. However, XCP requires the collaboration of all the routers on the data path which is almost impossible to achieve in an incremental deployment scenario of XCP. It has been shown that XCP behaves badly, worse than TCP, in the presence of non-XCP routers thus limiting dramatically the benefit of having XCP running in some parts of the network. In this paper, we address this problem and propose XCP-i which is operable on an internetwork consisting of XCP routers and traditional IP routers without loosing the benefit of the XCP control laws. The simulation results on a number of topologies that reflect the various scenario of incremental deployment on the Internet show that although XCP-i performances depend on available bandwidth estimation accuracy, XCP-i still outperforms TCP on high-speed links.

Programmable and active networks allow specified classes of users to deploy dynamic network services adapted to data streams requirements. Based on our experience in high performance active networking, this paper compares two alternative approaches for adding dynamic solutions in the network: heavy and lightweight dynamic network services. We propose and describe solutions to efficiently manage and deploy heavy services (requiring resources and closely linked with middleware or applications) and lightweight network services (generic and pragmatic solutions with limited impact on network infrastructure). Experiments on local and wide area platforms are presented.

The field of active and programmable networks is rapidly expanding. These networks allow users and network designers to easily deploy new services. Active networks must deal with the need of portability for services to be easily implemented in active routers and the need of high performance design to efficiently transport multimedia streams. While most of proposed systems deal with adaptability, flexibility and new protocols by using JAVA, only few systems focus on high performance. We explore the design of intelligent network by proposing the Tamanoir 1 system, a complete framework dedicated to high performance active networking, which allows users to easily deploy and maintain services on distributed active routers. To guarantee the portability, Tamanoir routers, services and tools are programmed in Java. But, high performance aspects are also taken into account by minimizing JAVA software latency with a compiled approach. First experiments and comparisons with other active networks Java-based systems will show the effectiveness and pertinence of our approach.

International audience