Overall Objectives
Research Program
Application Domains
New Software and Platforms
New Results
Bilateral Contracts and Grants with Industry
Partnerships and Cooperations
XML PDF e-pub
PDF e-Pub

Section: New Results

Green IT

Participants : Simon Dupont, Md Sabbir Hasan, Thomas Ledoux, Jonathan Lejeune, Guillaume Le Louët, Jean-Marc Menaud.

In 2015, we have provided new models and solutions for the energy-optimal execution of cloud applications in data centers.

Renewable energy

With the emergence of the Future Internet and the emergence of new IT models such as cloud computing, the usage of data centers (DC) and consequently their power consumption increase dramatically. Besides the ecological impact, the energy consumption is a predominant criteria for DC providers since it determines the daily cost of their infrastructure. As a consequence, power management becomes one of the main challenges for DC infrastructures and more generally for large-scale distributed systems.

Renewable energy for data centers

We have presented the EPOC project which focuses on optimizing the energy consumption of mono-site DCs connected to the regular electrical grid and to renewable energy sources [19] . A first challenge in this context consists in developing a (for users) transparent distributed system that enables energy-proportional computations from the system to service-oriented levels. The second challenge addresses the corresponding energy issues through collaborative measurements and energy-optimizing actions inside infrastructure-software stack, more precisely between applications and resource management systems. This approach must manage Service Level Agreement (SLA) constraints by striving for the best trade-off between energy cost (from the regular electric grid), its availability (from renewable energy sources), and service degradation (from application reconfiguration issues to job suspension ones). The third challenge embarks pursues energy efficient optical networks as key enablers of the future internet and cloud-networking service deployment through the convergence of optical infrastructure with the upper network layers.

The second challenge is more precisely describe in [30] . In this paper we present PIKA, a framework aiming at reducing the Brownian energy consumption (ie. from non renewable energy sources), and improving the usage of renewable energy for mono-site data centers. PIKA exploits jobs with slack periods, and executes and suspends them depending on the available renewable energy supply. By consolidating the virtual machines (VMs) on the physical servers, PIKA adjusts the number of powered-on servers in order for the overall energy consumption to match the renewable energy supply. Using simulations driven by real-world workloads and solar power traces, we demonstrate that PIKA consumes 41% less Brownian energy and increases 35.3% renewable energy integration ratio in comparison with the baseline algorithm from the literature.

Energy monotoring

We have designed SensorScript, a Business-Oriented Domain-Specific Language for Sensor Networks [24] , [35] . In smart grids, or more generally the Internet of Things, many research work has been performed on the whole chain, from communication sensors to big data management, through communication middlewares. Few of this work have addressed the problem of gathered data access. In fact, being able, as a system administrator, to manipulate and gather data collected from a set of sensors in a simple and efficient way represents an essential need.

To address this issue, the solution we considered consists of a multi-context modeling for raw data, in the form of a multi-tree: a directed acyclic graph consisting of multiple intricate trees, each of them describing a hierarchy corresponding to a given use context. The objectives are to provide not only a means to rationalize users needs before writing queries, but also to offer a domain-specific language (DSL) which takes advantage of the multi-tree modeling to simplify the experience of pre-identified users that query data.

Green SLA and virtualization of green energy

The demand for energy-efficient services is increasing considerably as people are getting more environmentally-conscious in order to build a sustainable society. The main challenge for Cloud providers is to manage Green SLA (Service Level Agreement) constraints for their customers while satisfying their business objectives, such as maximizing profits by lowering expenditure for so-called green (renewable) energy. Since, Green SLA needs to be proposed based on the presence of green energy, the intermittent nature of renewable sources makes it difficult to be achieved. In response, we propose a scheme for green energy management based on three contributions [15] : i) we introduce the concept of virtualization of green energy to address the uncertainty of green energy availability, ii) we extend the Cloud Service Level Agreement (CSLA) language to support Green SLA by introducing two new threshold parameters and iii) we introduce algorithms for Green SLA which leverage the concept of virtualization of green energy to provide interval-specific Green SLA. We have conducted experiments with real workload profiles from PlanetLab and server power model from SPECpower to demonstrate that Green SLA can be successfully established and satisfied without incurring higher cost.