Research

You can find here an overview of my research, the list of my PhD students, and a list and description of the projects I am/was involved in

Overview

An overview of my research work is available in my Habilitation (document in french) or in the associated slides (slides in french), defended on 12th Nov 2015

My focus is on Energy Efficiency in ICT and with ICT. I mainly focuses on HPC and large scale infrastructures (clusters, grids, clouds). I’m professor in Sepia Team in IRIT Lab

A video on my research topic

Research Interests

  • Energy efficient task placement (centralized and decentralized)
  • Mathematical models of energy consumption
  • Green middleware for cluster/grid/cloud
  • Linux kernel governor for reducing HPC energy consumption
  • Energy consumption reduction in hybrid (Cpu/Gpu) architecture

Phd Students

  • Maël Madon: Ordonnancement sous contraintes environnementales pour centres de données à grande échelle
  • Ngoc Tu Dinh: Walk-In: Mise à jour des hyperviseurs dans un datacenter
  • Manal Benaissa: Optimisation de l’ensemble du processus de dimensionnement informatique et électrique d’un datacenter vert, de taille moyenne, alimenté uniquement par des sources d’énergie renouvelables et en tenant compte de l’incertitude des conditions environnementales et externes.
  • Daniel Lezcano: Amélioration de la gestion d’énergie sur architecture ARM

Former Phd Students

  • Léo Grange: Datacenter management for on-site intermittent and uncertain renewable energy sources
  • Ines Decourchelle: Vers une meilleure utilisation des énergies renouvelables : application à des bâtiments scientifiques
  • Zong Yi Liu: Self-Adaptive Bandwidth Control for Balanced QoS and Energy Aware Optimization in Wireless Sensor Network
  • Leandro Fontoura Cupertino: Modeling the power consumption of computing systems and applications through machine learning techniques
  • Tom Guerout: Ordonnancement sous contraintes de qualité de service dans les clouds
  • Cheikhou Thiam: Anti load-balancing for energy-aware distributed scheduling of virtual machines
  • Damien Borgetto: Allocation et réallocation de services pour les économies d’énergie dans les clusters et les clouds

Collaboration with Phd Students

  • Violaine Villebonnet: Scheduling and Dynamic Provisioning for Energy Proportional Heterogeneous Infrastructures
  • Tsafack Chetsa Ghislain Landry: System Profiling and Green Capabilities for Large Scale and Distributed Infrastructures
  • Samuel Asferaw: Reducing the Impact of Crowd Geo-localization on Battery Life by Proposing Novel Information Distribution Algorithm
  • Wahi Narsisian: Energy optimization methodology for scientific e‐infrastructure providers

Projects

Exa-SofT The Exa-SofT project (in the NumPEx PEPR) aims at consolidating the exascale software ecosystem by providing a coherent, exascale-ready software stack featuring breakthrough research advances enabled by multidisciplinary collaborations between researchers.

CareCloud The CARECloud project (understanding, improving, reducing the environmental impacts of Cloud Computing) aims to drastically reduce the environmental impacts of cloud infrastructures

Delight The DELIGHT project aims to evaluate and reduce the energy consumption of federated learning using different levers (gradient compression, data summarization, speed-scaling, etc.). Given the heterogeneity of the data, another objective will be to study the process of bargaining and coalition formation among nodes in order to understand to what extent a node has an interest in collaborating with others. The techniques developed will be empirically validated on computer vision and NLP tasks using the Flower toolkit.

DataZero 2 Since few years, several companies and universities have investigated the possibilities of using renewable energies to decrease the ever-growing environmental impact of datacenters. Few research initiatives have paved the way to understand how a datacenter ONLY operated by renewable energies can work. Since 2015 and till 2019, DATAZERO investigates new ways to power and manage one Cloud data center using on-site renewable energy sources and storage elements (photovoltaic, wind turbines, fuel cells, batteries, supercap). DATAZERO2 (2020-2024) aims at improving the operation and design of cooperating datacenters operated only with renewable energies, in particular using disruptive approaches for uncertainty management, cooling and power distribution. The second objective is to propose a comprehensive knowledge base and toolbox for industry and researchers for zero-emission datacenters beyond state-of-the-art solutions.

Finished projects

Energumen Energy consumption has always been a concern in High Performance Computing (HPC) platforms. Today, it becomes even more critical due to the transition to the next generation extreme scale platforms. In ENERGUMEN, we propose two new complementary mechanisms for addressing the energy/performance trade-off: First, we will revisit the classical speed-scaling and power down mechanisms by using a malleable job model, which allows to shrink or stretch dynamically the execution times of the jobs according to the current energy profile. We will also study optimized policies for energy-aware data allocations at the software level. Both mechanisms involve the design of sophisticated models and methodologies for the efficient exploitation of the idle periods at running time. There are hard scientific and technological challenges to determine the best trade-off between both mechanisms. In the quest of extreme scale platforms where a trade-off between performance and energy consumption is necessary, the originality of ENERGUMEN is to revisit the principles of existing resource managers and to investigate new functionalities by harnessing applications’ malleability.

neOCampus Action de l’universite Paul Sabatier neOCampus will demonstrate the campus of the future: connected, innovative, smart and sustainable.

i-Nondations i-Nondations (e-flooding) is a french research project funded by the ANR . It aims to handle fast floods to improve crisis management and resiliency. Every year floods happen. Solutions exist for slow floods but fast ones are difficult to predict and to handle. This project aims to model fast floods in term of risk management and impact on the infrastructures using data collected by technological or human sensors. The project will integrate different technical expertise to handle fast floods in crisis management and resiliency. The project suggests managing three phases: before, during, and after a crisis in a loop of feedback coming from the autonomic field called MAPE-K loop. It is based on four steps : Monitoring, Analysis, Planning and Execution with a Knowledge database. The monitoring will detect events, Analysis will analyze the risks based on models of predictions; creation of metrics to measure the impact on infrastructures; and proposing functions to evaluate alternative trajectories. The Plan will decide how to handle the crisis in order to find the optimal on the alternatives and then actions will be executed. Two loops will be used : one for short term timescale and one for long term. The short term one will aim to handle the crisis while the long term one will aim to prevent other crisis. Both loops will interact through a learning process. The Execute module of the loops will not be adressed in this project.

COST IC-1305: Network for Sustainable Ultrascale Computing (NESUS) 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. The network will contribute 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.

DataZero : Since some years energy efficiency in datacenters is improving but the amount of electricity needed for operating those for hosting Cloud services is growing with the sizes of the infrastructure and the user demands. With renewable energies and the usage of direct current in datacenters, we believe that we can cope with this problem. On the one hand several efforts have been conducted at the computing level in datacenters (for service placement and scheduling) and, on the other hand, for energy provisioning partly with renewable energies, but without much interaction between these two efforts. DATAZERO will give consistent solutions for high availability of IT (Information Technologies) services, avoiding unnecessary redundancies, under the constraints of intermittent nature of electrical and services flows. We noticed that, despite a true interest towards pragmatic and market-based solutions, comprehensive studies are missing to cope with the robustness of a cloud infrastructure in the conditions of datacenters powered by renewable energy. The question we address in DATAZERO is: How to manage the electricity and the service flows in order to deliver services to customers in a robust and efficient manner within datacenters operated with several energy sources?

PIXAL (2015-2016) : A new innovative platform for next generation teaching with interactive lessons, automatic evaluation of practical sessions and a nation-wide positioning test for computing science.

CoolEmAll (2011-2014) FP7 European project CoolEmAll will provide means for a deep insight into the total energy consumption of data centres. To this end, the project will provide advanced planning and optimization tools for modular data centre environments. Once developed, these tools should help to minimize the energy consumption, and consequently the CO2 emissions of the IT infrastructure with related facilities.

SOP: think global Services for persOnal comPuter (2011-2014) The objective of this ANR project is to build a hybrid working model for consumers and professional computing, that takes benefit from a mix between local installation/management and remote access to software and services located in distant dedicated datacenters or even in other non-used consumers’ machines

EcoToul (2012-2013)~: This projects aims at providing information concerning mobility in Toulouse. Data from thousands of smartphones users from several means of transportation (foot, bicycle, cars, public transports,…) will be merged in a global information systems. This data will be used to compute a transportation status map of Toulouse. This map will be used to give time of transit of several means of transportation to users.

COST IC-0804: Energy efficiency in large scale distributed systems (2009-2013) The main objective of the Action is to foster original research initiatives addressing energy awareness/saving and to increase the overall impact of European research in the field of energy efficiency in distributed systems. The goal of the Action is to give coherence to the European research agenda in the field, by promoting coordination and encouraging discussions among the individual research groups, sharing of operational know-how (lessons-learned, problems found during practical energy measurements and estimates, ideas for real-world exploitation of energy aware techniques, etc.)

ANR-JST FP3C (Framework and Programming for Post Petascale Computing) : French-Japanese Project on Petaflop applications (2010-2013) The goal of the project is to address the development of runtime systems for scientific applications running on petaflopic machines. Our contribution resides in the analysis of the energy consumption and leverages existing on such machines to handle their electricity hungriness.

PIREGRID: A Pyrenean Operational Grid (2009-2012) PireGrid is an INTERREG IV A project which have two main objectives: Deployment of a production Grid Computing infrastructure in the regions of Aragon, Aquitaine and Midi-Pyrenees. To achieve successful cases in execution of applications from small and medium size companies of these regions, in order to demonstrate its usage of these emerging technologies to the companies.

GreenNet (2008-2009) ARC Inria: Power aware software frameworks for high performance data transport and computing in large scale distributed systems