7.1.1 Tansiv

• Name:
  Time-Accurate Network Simulation Interconnecting Vms
• Keywords:
  Operating system, Virtualization, Cloud, Simulation, Cybersecurity
• Functional Description:
  Tansiv: Time-Accurate Network Simulation Interconnecting Virtual machines (VMs). Tansiv is a novel way to run an unmodified distributed application on top of a simulated network in a time accurate and stealth way. To this aim, the VMs execution is coordinated (interrupted and restarted) in order to garantee accurate arrival and transfer of network packets while ensuring realistic time flow within the VMs. The project can leverage several frameworks for simulating the data (SimGrid or ns-3) and several virtualization solutions to encapsulate the application, intercept the network traffic and enforce the interruption decision (Qemu in emulation mode, KVM and Xen with hardware-accelerated virtualization). Tansiv can be used in various situations: malware analysis (e.g. to defeat malware evasion technique based on network timing measures) or analysis of an application on a geo-distributed context.
• Contact:
  Louis Rilling
• Partner:
  DGA-MI

7.1.2 SimGrid

• Keywords:
  Large-scale Emulators, Grid Computing, Distributed Applications
• Scientific Description:

  SimGrid is a toolkit that provides core functionalities for the simulation of distributed applications in heterogeneous distributed environments. The simulation engine uses algorithmic and implementation techniques toward the fast simulation of large systems on a single machine. The models are theoretically grounded and experimentally validated. The results are reproducible, enabling better scientific practices.

  Its models of networks, cpus and disks are adapted to (Data)Grids, P2P, Clouds, Clusters and HPC, allowing multi-domain studies. It can be used either to simulate algorithms and prototypes of applications, or to emulate real MPI applications through the virtualization of their communication, or to formally assess algorithms and applications that can run in the framework.

  The formal verification module explores all possible message interleavings in the application, searching for states violating the provided properties. We recently added the ability to assess liveness properties over arbitrary and legacy codes, thanks to a system-level introspection tool that provides a finely detailed view of the running application to the model checker. This can for example be leveraged to verify both safety or liveness properties, on arbitrary MPI code written in C/C++/Fortran.

• Functional Description:
  SimGrid is a simulation toolkit that provides core functionalities for the simulation of distributed applications in large scale heterogeneous distributed environments.
• Release Contributions:

  The Anne of Brittany release (she became a Duchess 536 years ago).

  * Various improvements and unification of the simulation APIs

  * MC: Enable the verification of Python programs, and of condvars and iprobe calls.

  * MC: Exhibit the critical transition when a failure is found.

  * (+ internal refactoring and bug fixes)

• News of the Year:

  There was one major release in 2024. On modeling aspects, we further improved and unified the relevant APIs. The MPI and S4U interfaces are now more consistent and integrated, and a new plugin was introduced to model disk arrays (JBODs). We also pursued our efforts to improve the overall framework, through bug fixes, code refactoring and other software quality improvements. In particular, interfaces that were deprecated since almost a decade were removed to ease the maintenance burden on our community.

  On the model checking front, we are still working hard to find our first “wild bug”, a.k.a. existing bug in an arbitrary piece of code while all bugs found so far with Mc SimGrid are bugs that we purposely added to a given code base. To that extend, our efforts span upon three axes. We first extended the programming model, to allow for the verification of more programs and hopefully of programs containing bugs. The most notable extensions are that the verified programs can now be written in Python, involve MPI_iprobe, or use condition variables. Mc SimGrid can now exhibit the critical transition when a failure is found. Before that transition, at least one exploration is correct, After it, all explorations are faulty. We added this feature to help us characterize whether the found issue is an application bug (and help us understand its root cause), or whether it is yet another bug in Mc SimGrid itself. Finally, we fixed dozens of bugs and vastly optimized the verification code to improve our chances to find a wild bug. Still, we did not find any such bug yet, so the chase continues.

• URL:
  https://­simgrid.­org/
• Contact:
  Martin Quinson
• Participants:
  Adrien Lebre, Anne-Cécile Orgerie, Arnaud Legrand, Augustin Degomme, Arnaud Giersch, Emmanuelle Saillard, Frédéric Suter, Jonathan Pastor, Martin Quinson, Samuel Thibault
• Partners:
  CNRS, ENS Rennes

7.1.3 EnOSlib

• Keywords:
  Distributed Applications, Distributed systems, Evaluation, Grid Computing, Cloud computing, Experimentation, Reproducibility, Linux, Virtualization
• Functional Description:

  EnOSlib is a library to help you with your distributed application experiments on bare-metal testbeds. The main parts of your experiment logic is made reusable by the following EnOSlib building blocks:

  - Reusable infrastructure configuration: The provider abstraction allows you to run your experiment on different environments (locally with Vagrant, Grid’5000, Chameleon, IoT-LAB and more) - Reusable software provisioning: In order to configure your nodes, EnOSlib exposes different APIs with different level of expressivity - Reusable services: Install common services such as Docker, monitoring stacks, network emulation... - Reusable experiment facilities: Tasks help you to iterate faster on your experimentation workflow

  EnOSlib is designed for experimentation purpose: benchmark in a controlled environment, academic validation …

• Release Contributions:

  To reduce dependencies, the default pip package no longer includes Jupyter support.

  Add support for Ansible 8, 9 and 10

• URL:
  https://­discovery.­gitlabpages.­inria.­fr/­enoslib/
• Publications:
  hal-01664515, hal-01689726
• Contact:
  Mathieu Simonin
• Participants:
  Alexis Bitaillou, Volodia Parol-Guarino, Thomas Badts, Matthieu Rakotojaona Rainimangavelo, Mathieu Simonin, Baptiste Jonglez, Marie Delavergne

8.1.1 Federated fog/cloud infrastructure design

Participants: Guillaume Pierre, Chih-Kai Huang.

The decentralization of computing infrastructures creates new opportunities for designing extremely large-scale infrastructures that are not under the control of a single operator (as all the current cloud providers propose), but rather organized as a federation of large numbers of smaller, local infrastructure operators [64]. Besides the obvious issue of establishing trust between independent operators (which may be addressed by technological or legal means), realizing this vision will require a major redesign of current infrastructure orchestrators. Instead of a monolithic platform composed of a small number of “control plane”' nodes which manage large numbers of “worker”' nodes, it becomes necessary to rethink our computing infrastructures as a horizontal collection of independent entities which may collaborate with each other while retaining local control over their resource management policies.

We proposed UnBound, a framework for meta-federations which addresses the difficult multi-tenancy issue in this context to isolate workloads belonging to different users and fog providers. Extensive evaluations with federations of up to 500 geo-distributed Kubernetes clusters demonstrate that UnBound maintains comparable application deployment times to the state of the art in a single member cluster scenario, avoids increasing cross-cluster network traffic, keeps resource consumption within acceptable limits, and exhibits stability and scalability, making it a suitable solution for large-scale fog computing deployments 13.

Monitoring large fog computing federations is also a challenge as standard solutions such as Prometheus Federations transmit detailed monitoring data about each cluster member at fixed intervals, which generates large network bandwidth usage. We proposed Acala, an aggregate monitoring framework for geo-distributed Kubernetes cluster federations which aims to provide the management cluster with aggregated information about the entire cluster instead of individual servers. Based on actual deployment under a controlled environment in the geo-distributed Grid'5000 testbed, our evaluations show that Acala reduces the cross-cluster network traffic by up to 97% and the scrape duration by up to 55% in the single member cluster experiment. Our solution also decreases cross-cluster network traffic by 95% and memory resource consumption by 83% in multiple member cluster scenarios 6.

These contributions and others from previous years are also reported in Chih-Kai Huang's PhD thesis 21.

8.1.2 Fog/cloud storage services

Participants: Shadi Ibrahim, Marc Tranzer, Mohammad Rizk.

Replication has been successfully deployed and practiced to ensure high data availability in large-scale distributed storage systems. However, with the relentless growth of generated and collected data, replication has become expensive not only in terms of storage cost, but also in terms of network cost, hardware cost and energy cost. While already popular for archived data in peer-to-peer (P2P) systems and cold data, in recent years, erasure coding (EC) has been progressively performed on the critical path of data access and has been has been increasingly used in practical storage systems such as Ceph and HDFS. The main reason for such an adoption is that EC can provide the same fault tolerance guarantee as replication, but at a much lower storage cost.

Storage devices account for nearly 11% of the total energy consumed by a data center, so using EC could result in significant energy savings. However, EC requires large data exchanges when reading data or recovering from failures, leading to higher energy consumption. Accordingly, and in an effort to understand the tradeoffs between data access performance, repair bandwidth and energy overhead, we conducted large-scale experiments on a CephFS cluster to evaluate the energy and performance of read, write and data repair under EC and replication 27. A more detailed article on the same topic is in preparation.

We also started working on the design and implementation of a scalable and reliable P2P storage system using EC as a fault tolerance technique. We have We implemented and evaluated the performance of EC in IPFS – an open source and is widely used P2P storage system – and show that our implementation can provide a good performance compared to replication when reading and writing data to IPFS 26. A more detailed article on the same topic is in preparation.

8.1.3 Reliable fog platforms in adverse natural environments

Participants: Guillaume Pierre, Ammar Kazem, Matthieu Nicolas.

Natural environment observatories allow a wide range of scientists such as biologists, botanists and hydrologists, to observe a zone of particular interest from the points of view of their different scientific disciplines. They follow a data-driven approach based on a variety of sensors and/or actuators deployed in the natural environment, coupled with different techniques to report the produced data to a public or private cloud for further analysis. The constraints which stem from the specificities of such observatories however deviate from traditional IoT use-cases deployed in urban environments. Observatories are often created in remote locations, where energy supply, cellular networking coverage and human maintenance are more challenging than usual. To address these challenges we are designing new fog computing platforms that can process observation data in-situ while respecting the specific constraints of these environments.

In this context we conducted a survey of 35 observatories in France and abroad to better understand their existing systems and practice as well as avenues for improvement using fog computing technologies 23. A more detailed article on the same topic is in preparation.

We also started addressing the difficult question of designing a fog computing cluster from a hardware and software point of view that has sufficient processing capacity to handle a pre-defined set of data processing applications while reducing its energy consumption as much as possible. For this we exploit heterogeneous clusters of single-board computers (Raspberry Pi or equivalent) that can offer interesting tradeoffs between processing power and necessary energy.

Finally we started to redesign and improve the already-existing LivingFog platform 2. The objective is to specialize it to become a viable platform to host a set of data processing applications in a natural environment observatory. The short-time goals are to help detect sensor failures in order to issue timely alarms. In a longe run we will extend the set of applications to include more specialized processing functions.

8.2.1 Reliable serverless runtimes

Participants: Nikos Parlavantzas, Mohamed Cherif Zouaoui Latreche.

Function-as-a-Service (FaaS) is a compelling programming model for developing applications that run on fog infrastructures. FaaS applications are composed of ephemeral, event-triggered functions, which can be flexibly deployed at any location on the Edge-Cloud continuum, can be rapidly triggered, and consume resources only when needed. However, supporting fog applications places stringent requirements on FaaS runtimes. First, the runtimes should support latency objectives for functions, critical for latency-sensitive fog applications. Second, the runtimes should support reducing the energy consumption of the underlying infrastructure, particularly important for resource-constrained fog nodes. A major limitation of current FaaS runtimes is their lack of support for meeting latency and energy consumption requirements.

To address this limitation, we proposed an approach for FaaS scheduling, called FoRLess 14. The approach uses Deep Reinforcement Learning, specifically Deep Q-Networks (DQN), to learn how to schedule functions in order to jointly optimize latency and energy consumption. We implemented the approach by extending Fission, a popular Kubernetes-based open-source FaaS runtime, and evaluated its effectiveness through experiments on the Grid’5000 testbed. The experimental results demonstrated that FoRLess significantly outperforms baseline scheduling methods in both latency and energy efficiency. This research is part of the thesis project of Mohamed Cherif Zouaoui Latreche, started in October 2023, in collaboration with Hector Duran-Limon of the University of Guadalajara, Mexico.

8.2.2 Dynamic platform adaptation for Federated Learning

Participants: Cédric Tedeschi, Mathis Valli.

Federated Learning (FL) has emerged as a paradigm shift enabling heterogeneous clients and devices to collaborate on training a shared global model while preserving the privacy of their local data. However, a common yet impractical assumption in existing FL approaches is that the deployment environment is static, which is rarely true in heterogeneous and highly-volatile environments like the Edge-Cloud Continuum, where FL is typically executed. While most of the current FL approaches process data in an online fashion, and are therefore adaptive by nature, they only support adaptation at the ML/DL level (e.g., through continual learning to tackle data and concept drift), putting aside the effects of system variance. Moreover, the study and validation of FL approaches strongly rely on simulations, which, although informative, tends to overlook the real-world complexities and dynamics of actual deployments, in particular with respect to changing network conditions, varying client resources, and security threats.

In a paper published at the FlexScience workshop 17, we made a first step to address these challenges. We investigated the shortcomings of traditional, static FL models and identified areas of adaptation to tackle real-life deployment challenges. We devised a set of design principles for FL systems that can smartly adjust their strategies for aggregation, communication, privacy, and security in response to changing system conditions. To illustrate the benefits envisioned by these strategies, we presented the results of a set of initial experiments on a 25-node testbed. The experiments, which vary both the number of participating clients and the network conditions, show how existing FL systems are strongly affected by changes in their operational environment. Based on these insights, we proposed a set of take-aways for the FL community, towards further research into FL systems that are not only accurate and scalable but also able to dynamically adapt to the real-world deployment unpredictability.

8.2.3 Reliable data stream processing runtimes

Participants: Guillaume Pierre, Shadi Ibrahim, Khaled Arsalane, Govind Kovilkkatt Panickerveetil.

Although data stream processing platforms such as Apache Flink are widely recognized as an interesting paradigm to process IoT data in fog computing platforms, the existing performance models that capture stream processing in geo-distributed environments are theoretical works only, and have not been validated against empirical measurements.

In this context we started exploring the limitations of stream processing autoscaling systems when dealing with stateful data processing operators in geo-distributed environments 8. We demonstrate that Flink's backpressure mechanism should not be used as the only trigger for rescaling operations in heterogeneous network conditions. Raw performance, as well as performance predictability, also degrade quickly in the presence of stateful data processing operators and/or high network latency between the processing nodes. This work is still ongoing and a second publication on this topic is in preparation.

Another issue related to data stream processing focuses on the message brokers such as Apache Kafka that are used to ingest incoming data and reliably distribute them to a data stream processing engine. It is very difficult to right-size such systems because their performance and resource efficiency varies a lot based on many hardware- and configuration-related aspects. Because of their stateful nature they are also difficult to dynamically rescale after their initial deployment, which leads many users to drastically over-provision them “just in case.” We are currently working on a methodology to better assist Kafka users in exploring the space of workable configurations and in selecting the optimal one for a given expected workload. A publication on this topic is in preparation.

8.2.4 Decentralized resource provisioning for serverless applications

Participants: Nikos Parlavantzas, Volodia Parol-Guarino.

Application resource provisioning in decentralised environments poses significant challenges. First, resources are typically owned and managed by multiple, independent providers, ranging from individuals and small-scale edge cloud operators to large, traditional cloud providers. All these providers should be incentivized to make their resources available to applications. Second, resources are highly heterogeneous and volatile, which, combined with the dynamism of application workloads, makes it difficult to meet application quality of service (QoS) requirements.

We aim to address these challenges in the context of FaaS applications. To this end, we have proposed a market-based approach, called GIRAFF, for placing FaaS applications in multi-provider fog infrastructures, along with an open-source implementation. In our approach, clients submit application placement requests associated with SLAs that specify guarantees over network latency and allocated resources. A marketplace then organizes an auction where fog nodes bid on the SLA to determine the nodes that will host the functions and the revenue for fog node owners.

In 2024, we extended GIRAFF to add support for function chains, proposed a fog-specific cost model, and performed reproducible experiments on the Grid’5000 testbed at large scale (up to 663 fog nodes on 60 servers) in order to compare GIRAFF with baseline placement methods. The evaluation demonstrates GIRAFF’s effectiveness in optimizing FaaS application placement. An article about these results is under submission. This research is part of the thesis project of Volodia Parol-Guarino, started in October 2022.

8.2.5 Data processing at scale

Participants: Shadi Ibrahim.

In computing continuum workflow, the data is processed in stages, with the completion of the later stages being highly dependent on the completion of the previous stage. Within each stage, there are tasks that run in parallel and perform complex computations. The completion time of a stage is highly dependent on the completion of the last task within that stage. In an attempt to optimize the data-intensive workflows in the computing continuum, this work investigates the factors that can affect performance within and between stages for future optimization strategies 24. In addition, an HPC performance analysis methodology is adapted based on our findings to validate that our results can be applied across platforms. A more detailed article on the same topic is in preparation.

Deep Neural Network (DNN) inference has been widely adopted in today’s intelligent applications, such as autonomous driving, virtual reality, image recognition. Usually, inference services are implemented as a workflow consisting of multiple DNN models. As the demand for parallel computing in DNN models is continuously growing, inference computations are gradually moving to GPUs. Serverless inference (i.e., deploying inference workflows on serverless computing) is gaining increasing popularity for proven high elasticity, cost efficiency (i.e., pay-as-you-go), and transparent deployment. In serverless inference, each DNN model is encapsulated as a "function", which can automatically scale up or down according to the change in number of requests. Existing serverless inference systems isolate functions in separate monolithic GPU runtimes (e.g., CUDA context), which is too heavy for short-lived and fine-grained functions, leading to a high startup latency, a large memory footprint, and expensive inter-function communication. Accordingly, we propose a new lightweight GPU sandbox for serverless inference workflow, called StreamBox 18. StreamBox unleashes the potential of streams and efficiently realizes them for serverless inference by implementing fine-grain and auto-scaling memory management, allowing transparent and efficient intra-GPU communication across functions, and enabling PCIe bandwidth sharing among concurrent streams. Evaluations over real-world workloads show that StreamBox reduces the GPU memory footprint by up to 82% and improves throughput by 6.7X compared to state-of-the-art serverless inference systems.

8.2.6 Mitigation of tail processing latency

Participants: Shadi Ibrahim.

The large scale and heterogeneity of distributed processing infrastructures often result in partial failures and unavoidable performance variability. This, in turn, causes unexpected long tails in job execution and the appearance of stragglers (i.e., tasks that take significantly longer time to finish than the normal execution time). Big data analytic systems employ speculative execution to mitigate stragglers at large scale. Specifically, new copies of detected stragglers are launched on available machines. Straggler detection plays an important role in the effectiveness of speculative execution. The methods used to detect stragglers use the information extracted from the last received heartbeats which may be outdated when the detection is triggered. In this ongoing work 22, we investigate how heartbeat arrivals and the use of outdated information in straggler detection affect the accuracy of straggler detection. Our preliminary results show that considering heartbeat timestamps can improve the accuracy of straggler detection, especially when combined with rate based detection methods which typically mark as stragglers tasks whose estimated elapsed times are 20% longer than the mean or median.

8.3.1 Energy accounting of digital services

Participants: Anne-Cécile Orgerie, Maxime Agusti, Vladimir Ostapenco.

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. Existing energy models are generally based on system usage data, which is incompatible with the general privacy policies of bare-metal server contracts, as proposed by OVHCloud. To deal with these problems, we proposed an original non-intrusive method for estimating the energy consumption of a server cooled by direct-chip liquid-cooling, based on the coolant temperature and the processor temperature obtained via IPMI 7. Our approach, developed in collaboration with OVHCloud, is evaluated on an experiment carried out on 19 bare metal servers of a production infrastructure equipped with physical wattmeters.

Data centers are very energy-intensive facilities whose power provision is challenging and constrained by power bounds. In modern data centers, servers account for a significant portion of the total power consumption. In this context, the ability to limit the instant power consumption of an individual computing node is an important requirement. There are several energy and power capping techniques that can be used to limit compute node power consumption, such as Intel RAPL. Although it is nowadays mainly utilized for energy measurement, Intel RAPL (Running Average Power Limit) was originally designed for power limitation purposes. It is unclear exactly how Intel RAPL technology operates and what effects it has on application performance and power consumption. We conducted a thorough analysis of Intel RAPL technology as a power capping leverage on a variety of heterogeneous nodes for a selection of CPU and memory intensive workloads 15. For this purpose, we first validated Intel RAPL power capping mechanism using a high-precision external power meter and investigated properties such as accuracy, power limit granularity, and settling time. Then, we attempted to determine which mechanisms are employed by RAPL to adjust power consumption.

8.3.2 Water consumption of Cloud data centers

Participants: Anne-Cécile Orgerie, Emmanuel Wedan Gnibga.

In an era of growing water demand and amplifying climatic disruptions, water is an increasingly limited and precious resource. As data centers continue to proliferate due in particular to growing artificial intelligence demand, datacenter reliance on water for cooling is drawing increasing criticism. In this work, we investigated the water and energy consumption, carbon emissions, and cost associated with the energy demand (including its cooling) from a datacenter, in four diverse locations (California, Texas, Germany, and France) 12. We considered water use both in evaporative cooling and power generation, employing a hybrid datacenter cooling model that is operable with and without evaporative cooling. Subsequently, we examined solutions that expand the equipment’s operational range and actively exploit the cooling design redundancy. Moreover, we investigated the benefits of making data centers dynamic, that is changing operating balance hourly.

Going further, we looked at the environmental impact of digital infrastructures through the particular prism of water-related issues in the data centres (DCs) that make up the clouds 4. We attempted to draw up an overview of the use of water in DCs, whether this use is direct, for cooling and humidifying electronic equipment, or indirect, for electricity generation, production and the end-of-life of equipment. This overview is illustrated with figures, mainly from the cloud service providers themselves, to give some orders of magnitude on water consumption and trends. The aim is also to discuss these figures, the relevance of the usual indicators and the associated potential impact transfers, and to recall the major spatiotemporal issues and conflicts of use linked to water resources.

8.3.3 Steering efficiency for distributed infrastructures

Participants: Anne-Cécile Orgerie, Martin Quinson, Emmanuel Wedan Gnibga, Matthieu Silard.

Integrating larger shares of renewables in data centers' electrical mix is mandatory to reduce their carbon footprint. However, as they are intermittent and fluctuating, renewable energies alone cannot provide a 24/7 supply and should be combined with a secondary source. Finding the optimal infrastructure configuration for both renewable production and financial costs remains difficult. In this work, we examined three scenarios with on-site renewable energy sources combined respectively with the electrical grid, batteries alone and batteries with hydrogen storage systems. The objectives were first, to size optimally the electric infrastructure using combinations of standard microgrids approaches, secondly to quantify the level of grid utilization when data centers consume/export electricity from/to the grid, to determine the level of effort required from the grid operator, and finally to analyze the cost of 100% autonomy provided by the battery-based configurations and to discuss their economic viability 5.

Self-consumption is developing worldwide to increase renewable electricity consumption and reduce electricity bills. It can be carried out individually or collectively (grouping several entities), but is generally restricted geographically. In datacenters, one may use load shifting to benefit from the self-consumption tariffs, at the cost of increasing energy consumption. An alternative would be to extend the self-consumption rules to wider perimeters. In this work, we proposed a comparative study on several aspects influencing the collective self-consumption (CSC) of an Edge infrastructure, including spatial load shifting, temporal load shifting and extending the current rules to encompass wider geographical boundaries 11.

Distributed infrastructures also include other systems than ICT. In particular, we are interested in smart grids that allow to efficiently perform demand-side management in electrical grids in order to increase the integration of fluctuating and/or intermittent renewable energy sources in the energy mix. In this work, we consider the computing infrastructure that controls the smart grid. In particular, in the context of the EDEN4SG ANR project, we consider the wide-scale deployment of electrical vehicles. Driven by governmental policies, particularly in Europe, Electric Vehicles (EVs) are gradually replacing the current fleet. Since they are only used during a small portion of the time for transportation, their batteries can be used for other purposes during the rest of the time. The power grid constantly requires to maintain the balance between electricity production and consumption. Controlling the charge of EVs can avoid the need to invest in grid reinforcement. This research work focuses on flattening the consumption in residential areas, also known as peak shaving, which reduces the need for reactive power plants on the grid. Existing solutions often rely on coordination among vehicles, meaning that they require a communication network between the various consumers in the area (e.g., vehicles, houses). However, such a dedicated communication network is not yet deployed in practice. It would be costly to deploy and operate, and could potentially constitute a privacy breach for consumers. In this work, we proposed a local smart charging solution only relying on information exchanges between each house's smartmeter and its EV's charging station 16. We compared our local peak shaving to solutions without communication infrastructure, and to an optimal theoretical solution provided by a linear programming solver. Simulation campaigns show that our local peak shaving algorithm offers a solution close to the optimum, with advantages in terms of voltage stability, fairness among users and vehicle battery lifetime preservation.

8.3.4 Frugal mobile computing

Participants: Martin Quinson, Victorien Elvinger.

Mobile computing is very representative of the ever increasing computational offering. In the recent decades, the hardware capacity has increased exponentially in both computational power, memory and storage space, and display size. This increase is in pace with the software offering that went from scarce text messages to augmented reality and advanced features leveraging embedded neural networks. The only decreasing metric is maybe the battery which was reduced from one week or more to one day or less, despite the steady increase of the battery capacity.

In this context, the SmolPhone project is research action exploring other potential designs and their respective advantages for mobile computing. Its practical aspects consist in designing a sort of low-tech smartphone offering some services of a classical smartphone with a one-week battery lifetime. The goal is not to optimize a typical smartphone but rather to reconsider the design space of mobile computing and beyond. We envision the designed solution not as a marketable product but as an effective workbench for future research in the domain of frugal computing.

The AEx SmolPhone started this year. On the hardware side, we evaluated the practical feasibility of the SmolPhone vision, which involves heterogeneous multiprocessors and low-power technologies for the screen and network. We identified several usage modes according to the provided features, and quantified the energy consumption of the processing units in each mode, as well as the time required to switch between modes 25. A similar evaluation of the broadband network energy consumption is still to be done in this context.

On the software side, our work on the evaluation of alternate solutions to browse the web on a device too limited to display complex HTML5 pages is still ongoing. We are also currently exploring the design of an applicative framework enabling one to easily develop graphical applications targeting such hardware.

8.4.1 Understanding software infrastructures

Participants: Léo Cosseron, Martin Quinson, Matthieu Simonin.

One of the most challenging infrastructure to study is the one deployed by malware operators. The difficulties classically posed by distributed infrastructures add to the fact that the infrastructure conceptors explicitly try to evade the analysis and/or attack the analysis system. As an answer, Virtual Machine Introspection (VMI) is used by sandbox-based analysis frameworks to observe malware samples while staying isolated and stealthy. Sandbox detection and evasion techniques based on hypervisor introspection are becoming less of an issue since running server and workstation environments on hypervisors is becoming standard. However, the fake network environment around a sandbox VM offers opportunities similar to hypervisor introspection for malware to evade. Malware can evaluate the discrepancy between observed performances and a real, presumed network environment of infected targets. VMI pauses also cause visible network performance glitches. To solve these issues, we proposed to extend virtual clock manipulation to synchronize hardware-accelerated virtual machines with a discrete-event network simulator 9.

8.4.2 Predictive models and simulation frameworks

Participants: François Lemercier, Anne-Cécile Orgerie, Martin Quinson, Clément Courageux-Sudan.

The deployment of applications closer to end-users through fog computing has shown promise in improving network communication times and reducing contention. However, the use of fog applications such as microservices necessitates intricate network interactions among heterogeneous devices. Consequently, understanding the impact of different application and infrastructure parameters on performance becomes crucial. Current literature either offers end-to-end models that lack granularity and validation or fine-grained models that only consider a portion of the infrastructure. We first compared experimentally the accuracy of the existing simulation frameworks. Then, we used SimGrid to obtain comprehensive metrics regarding microservice applications operating in the fog, both for performance evaluation (e.g. application latency) and environmental evaluation (e.g. energy consumption and greenhouse gas emissions) 10.

8.4.3 Sustainable research ecosystem

Participants: Anne-Cécile Orgerie, Martin Quinson, Matthieu Simonin.

Scientific computing platforms offer computing and storage resources that are shared between multiple users. They generally do not have access to information that would enable them to assess the environmental footprint of their use of these resources. However, calculating their footprint could raise awareness of the environmental impacts generated by the use of these platforms and, ideally, lead to the organisation of a reduction in this footprint. The environmental footprint is often reduced to measuring its carbon footprint, or even the carbon emissions linked to its use alone. However, the reality of environmental impacts is much more complex, being multi-factorial and exacerbated by technical acceleration and intensification of use. As a complete analysis of environmental impacts is hampered by the lack of data that can be systematically obtained throughout the life cycle of digital devices, our study nevertheless focuses on a practical case: the calculation of carbon costs in a computing centre dedicated to research (the Rennes site of Grid'5000) 19. We first explored the various methods of calculating the environmental footprint that can be applied to the management of a computing infrastructure, including embodied energy; and secondly we presented and discussed a calculation method that can be used to attribute a carbon cost to the use of a computing core for one hour.

Défi Inria OVHCloud

Participants: Anne-Cécile Orgerie, Shadi Ibrahim, Govind Kovilkkatt Panickerveetil, Guillaume Pierre, Maxime Agusti, Vladimir Ostapenco, Marc Tranzer.

• Title:
  FrugalCloud: Eco-conception de bout en bout d'un cloud pour en réduire les impacts environnementaux
• Partner Institution(s):
   

  • Inria, France
  • OVHCloud, France
• Date/Duration:
  2021-2025
• Additionnal info/keywords:
  The goal of this collaborative framework between the OVHcloud and Inria is to explore new solutions for the design of cloud computing services that are more energy-efficient and environment friendly. Five Inria project-teams are involved in this challenge including Avalon, Inocs, Magellan, Spirals, Stack. Members of the Magellan team contribute to four sub-challenges including (1) Software ecodesign of a data stream processing service; (2) energy-efficient data management; (3) observation of bare metal co-location platforms and proposal of energy reduction catalogues and models; and (4) modelling and designing a framework and its environmental Gantt Chart to manage physical and logical levers.

Défi Inria Hive

Participants: Shadi Ibrahim, Mohammad Rizk, Quentin Acher.

• Title:
  Alvearium: Large Scale Secure and Reliable Peer-to-Peer Cloud Storage.
• Partner Institution(s):
   

  • Inria, France
  • Hive, France
• Date/Duration:
  2023-2026
• Additionnal info/keywords:
  The goal of this collaborative framework between Hive and Inria is to explore new solutions for the design and realization of large scale secure and reliable Peer-to-Peer Cloud storage. Four Inria project-teams are involved in this challenge including COAST, Magellan, WIDE, COATI. Members of the Myriads team will contribute to two axes. Specifically, the Magellan team coordinates the axis on reliable and cost-efficient data placement and repair in P2P storage over immutable data; and contributes to the axis on the management of mutable data over P2P storage.

10.1.1 Associate Teams in the framework of an Inria International Lab or in the framework of an Inria International Program

10.2.1 Visits of international scientists

10.3.1 Digital Europe

AEx Smolphone

Participants: Martin Quinson, Victorien Elvinger.

• Title:
  SmolPhone: un smartphone conscient des limites énergétiques.
• Partner Institution(s):
   

  • Inria Magellan team
  • Inria Taran team
• Date/Duration:
  2024-2026
• Additionnal info/keywords:
  The SmolPhone project is research action exploring potential frugal and low-tech solutions in the domain of mobile computing. Its practical aspects consist in designing a sort of low-tech smartphone offering some services of a classical smartphone with a one-week battery lifetime. The goal of the AEx project is to unlock several technical difficulties by conceiving a software and hardware development workbench enabling future research on the topic of frugal computing and low-tech IT services.

ADT SmartObs

Participants: Guillaume Pierre, Matthieu Nicolas.

• Title:
  Une plateforme originale pour le monitoring environemental
• Partner Institution(s):
   

  • Inria Magellan team
  • Inria Taran team
• Date/Duration:
  2024-2025
• Additionnal info/keywords:
  The goal of the SmartObs project is to help develop the SmartSense and LivingFog platforms dedicated to natural environment monitoring, and to enable interdisciplinary collaborations with partners from the environmental sciences domain. The objective is to generate, analyse and transmit smart sensor data (SmartSense) and further process these data in situ (LivingFog) before sending them to a remote cloud.

ANR Dark-Era

Participants: Martin Quinson, .

• Title:
  Dark-Era: Dataflow Algorithm aRchitecture co-design of SKA pipeline for Exascale RadioAstronomy
• Partner Institution(s):
   

  • CentraleSupélec, Laboratoire des Signaux et Systèmes (LSN)
  • INSA Rennes, Institut d'Electronique et de Télécommunication de Rennes (IETR)
  • Observatoire Paris, Galaxies, Etoiles, Physique, Instrumentation
  • Observatoire de la Côte d'Azur Nice, Laboratoire J-L. Lagrange
  • ENS Rennes, Institut de Recherche en Informatique et Systèmes Aléatoires (IRISA)
• Date/Duration:
  2021-2025
• Additionnal info/keywords:
  The future Square Kilometer Array (SKA) radio telescope poses unprecedented challenges to the underlying computational system. The instrument is expected to produce a sustained rate of Terabytes of data per second, mandating on-site pre-processing to reduce the size of data to be transferred. However, the electromagnetic noise of a traditional computing center would hinder the quality of the measurements if located near to the instrument. As a result, the Science Data Processor (SDP) pipeline will only have an energy budget of only 1 MWatt to execute a complex algorithm chain estimated at 250 Petaops/s. Because of these requirements, the SDP must be an innovative data-oriented infrastructure running on a disaggregated architecture combining standard HPC systems with dedicated accelerators such as GPU or FPGA. The goal of the DarkEra project is to contribute to the performance assessment both in time and energy of new complex scientific algorithms on not-yet-existing complex computing infrastructures. To that extend, a prototyping tool will be developed for the prospective profiling of data-oriented applications during their development.

ANR FACTO

Participants: Anne-Cécile Orgerie, François Lemercier.

• Title:
  FACTO: A Multi-Purpose Wi-Fi Network for a Low-Consumption Smart Home
• Partner Institution(s):
   

  • CNRS, IRISA lab
  • University of Lyon 1, LIP lab
  • Orange
  • Fondation Blaise Pascal
• Date/Duration:
  2021-2024
• Additionnal info/keywords:
  The number of smart homes is rapidly expanding worldwide with an increasing amount of wireless IT devices. The diversity of these devices is accompanied by the development of multiple wireless protocols and technologies that aim to connect them. However, these technologies offer overlapping capabilities. This overprovisioning is highly suboptimal from an energy point of view and can be viewed as a first barrier towards sustainable smart homes. Therefore, in the FACTO project, we propose to design a multi-purpose network based on a single optimized technology (namely Wi-Fi), in order to offer an energy-efficient, adaptable and integrated connectivity to all smart home's devices.

ANR EDEN4SG

Participants: Anne-Cécile Orgerie, Matthieu Silard.

• Title:
  EDEN4SG: Efficient and Dynamic ENergy Management for Large-Scale Smart Grids
• Partner Institution(s):
   

  • CNRS, IRISA lab
  • CNRS, SATIE lab
  • University of Toulouse, IRIT lab
  • EDF
  • SRD Energies
  • Orange
• Date/Duration:
  2023-2027
• Additionnal info/keywords:
  Climate change as well as geopolitical tensions have led a large number of countries to target a massive integration of renewables in their energy mix. This will be achieved among others by increasing the electrification rate of several sectors such as transport. In this context, the wide-scale deployment of electrical vehicles (EVs) represents a challenge as well as an opportunity to render more efficient and affordable the transformation of the current power system into a smarter grid. The project targets to develop methods for the intelligent coordination of large-scale EV fleets and as well to determine the associated cost of information for piloting the required smart grid.

IPCEI-CIS E2CC (BPI)

Participants: Anne-Cécile Orgerie.

• Title:
  Eco Edge to Cloud Continuum
• Partner Institution(s):
   

  • CNRS, IRISA lab
  • Inria, LIG lab
  • University of Toulouse, IRIT lab
  • CNRS, LAAS lab
  • ATOS (Bull SAS)
  • Armadillo
  • CGI France
  • NBS
  • Ningaloo
  • Provenrun
  • Ryax
• Date/Duration:
  2024-2028
• Additionnal info/keywords:
  The aim of the IPCEI-CIS E2CC project is to provide a standardized end-to-end platform that can be connected natively to any cloud provider, offering the right level of interoperability and service portability and creating a continuum from Edge to Cloud, including cybersecurity, decarbonisation, orchestration and platform functions. This platform will provide hardware and software solutions, exhibiting vertical services (computer vision, decarbonisation services, etc.), cybersecurity services, AI/MLOPs, and Baremetal & Edge infrastructures.

CARECloud (PEPR Cloud)

Participants: Anne-Cécile Orgerie, Matteo Chancerel, Pablo Leboulanger, Thomas Stavis.

• Title:
  CARECLOUD: Understanding, improving, reducing the environmental impacts of Cloud Computing.
• Partner Institution(s):
   

  • CNRS, IRISA lab
  • Inria, CRIStAL lab
  • CNRS, I3S lab
  • University of Toulouse, IRIT lab
  • Inria, LIP lab
  • IMT, LS2N lab
  • IMT, SAMOVAR lab
• Date/Duration:
  2023-2030
• Additionnal info/keywords:
  Cloud computing and its many variations offer users considerable computing and storage capacities. The maturity of virtualization techniques has enabled the emergence of complex virtualized infrastructures, capable of rapidly deploying and reconfiguring virtual and elastic resources in increasingly distributed infrastructures. This resource management, transparent to users, gives the illusion of access to flexible, unlimited and almost immaterial resources. However, the power consumption of these clouds is very real and worrying, as are their overall greenhouse gas (GHG) emissions and the consumption of critical raw materials used in their manufacture. In a context where climate change is becoming more visible and impressive every year, with serious consequences for people and the planet, all sectors (transport, building, agriculture, industry, etc.) must contribute to the effort to reduce GHG emissions. Despite their ability to optimize processes in other sectors (transport, energy, agriculture), clouds are not immune to this observation: the increasing slope of their greenhouse gas emissions must be reversed, otherwise their potential benefits in other sectors will be erased. This is why the CARECloud project (understanding, improving, reducing the environmental impacts of Cloud Computing) aims to drastically reduce the environmental impacts of cloud infrastructures.

CNRS DISTANT

Participants: Guillaume Pierre, Ammar Kazem, Matthieu Nicolas.

• Title:
  Automated Data qualIty asSurance for a criTical zone observAtory iNThe Himalayas – Kaligandaki River Nepal
• Partner Institution(s):
   

  • Université de Rennes – Géosciences Rennes, France
  • Université de Rennes – IRISA, France
  • IRD, France
  • German Research Centre for Geosciences GFZ, Germany
  • Tribhuvan University, Nepal
  • Smartphones4Water Citizen Science NGO, Nepal
• Date/Duration:
  2024-2025
• Additionnal info/keywords:
  Mountain landscapes respond disproportionately quickly to changes in external forcing by tectonics and climate. Erosion and associated extreme events are common and present a continuous threat to local populations. At the same time, mountains act as resources and supply  1/3 of the world population with fresh water. Yet mountains are notoriously under studied and are widely underrepresented in global monitoring networks. One particular problem of monitoring such landscapes is the data quality, especially in remote areas where manned support is difficult to maintain. The dedicated monitoring area is the Kaligandaki Catchment in Nepal, spanning an ideal transect across the Himalayan Mountain Range from the edge of the Tibetan Plateau, through the high mountains, to the low-lying Gangetic foreland. We will implement a LivingFog platform computing and data logging system that can detect bogus data and issue alerts to the operators. This system is designed to serve as a near real-time data processing system for early warning of catastrophic events in the future.

NF-JEN (PEPR 5G and future networks)

Participants: Anne-Cécile Orgerie.

• Title:
  NF-JEN: Just Enough Networks
• Partner Institution(s):
   

  • CNRS, IRISA lab
  • CEA, LETI
  • INSA de Lyon, CITI lab
  • Inria, CRIStAL lab
  • IMT, IEMN lab
  • CNRS, IETR lab
  • INP Bordeaux, IMS lab
  • IMT, Lab-STICC lab
  • ESIEE - Université Gustave Eiffel, LIGM lab
  • Inria, LIP lab
  • IMT, LTCI lab
  • CNRS, XLIM lab
• Date/Duration:
  2023-2027
• Additionnal info/keywords:
  Communication networks are often presented as a necessary means of reducing the impact on the environment of various sectors of industry. In practice, the roll-out of new generations of mobile broadband networks has required increased communication resources for wireless access networks. This has proved an effective approach in terms of performance but concerns remain about its energy cost and more generally its environmental impacts. Exposure to electromagnetic fields also remains a cause of concern despite existing protection limits. In the JEN (Just Enough Networks) project, we propose to develop just enough networks: network whose dimension, performance, resource usage and energy consumption are just enough to satisfy users needs. Along with designing energy-efficient and sober networks, we will provide multi-indicators models that could help policy-makers and inform the public debate.

STEEL (PEPR Cloud)

Participants: Cédric Tedeschi, Mathis Valli.

• Title:
  STEEL: Secure and Efficient Data Storage and Processing on Cloud-based Infrastructures
• Partner Institution(s):
   

  • DRIM (CNRS, INSA Lyon)
  • ERODS (Université Grenoble Alpes)
  • Kerdata (Inria, INSA Rennes)
  • IN2P3 (CNRS)
  • MAGELLAN (Université de Rennes, Inria)
  • PROGRESS (Université de Bordeaux, INP)
  • STACK (IMT Atlantique, Inria, LS2N)
  • TeraLab (IMT)
• Date/Duration:
  7 years
• Additionnal info/keywords:
  The strong development of cloud computing and its massive adoption for the storage of unprecedented volumes of data in a growing number of domains has brought to light major technological challenges. In this project we will address several of these challenges, organized in three research directions:

  • Exploitation of emerging technologies for efficient storage on cloud infrastructures. We will address this challenge through NVRAM-based distributed performance storage solutions, as close as possible to data production and consumption locations (disaggregation principle) and develop strategies to optimize the trade-off between data consistency and access performance.
  • Efficient storage and processing of data on hybrid, heterogeneous infrastructures within the digital edge-cloud-supercomputer continuum. The execution of hybrid workflows combining simulations, analysis of sensor data flows and machine learning requires storage resources ranging from the edge to cloud infrastructures, and even to supercomputers, which poses challenges for unified data storage and processing.
  • Confidential storage, in connection with the need to store and analyze large volumes of data of strategic interest or of a personal nature.

Taranis (PEPR Cloud)

Participants: Shadi Ibrahim, Nikos Parlavantzas.

• Title:
  Taranis: Model, Deploy, Orchestrate, and Optimize Cloud Applications and Infrastructure
• Partner Institution(s):
   

  • IMT, France
  • Université Grenoble Alpes, France
  • Université de Rennes, France
  • Inria, France
  • CNRS, France
  • CEA, France
  • ENS Lyon, France
  • Université Claude Bernard Lyon 1, France
  • Université de Lille, France
  • INSA Rennes, France
• Date/Duration:
  7 years
• Additionnal info/keywords:
  New infrastructures, such as Edge Computing or the Cloud-Edge-IoT computing continuum, complicate the cloud landscape as they add new challenges related to resource diversity and heterogeneity (from small sensors to data centers/HPC, from low power networks to core networks), geographical distribution, as well as increased dynamicity and security needs, all under energy consumption and regulatory constraints. In order to efficiently exploit new infrastructures, we propose a strategy based on a significant abstraction of the application structure description to further automate application and infrastructure management. Thus, it will be possible to globally optimize used resources with respect to multi-criteria objectives (price, deadline, performance, energy, etc.) on both the user side (applications) and the provider side (infrastructures). This abstraction also includes the challenges related to facilitating application reconfiguration and to automatically adapting the use of resources. The Taranis project addresses these issues through four scientific work packages, each focusing on a phase of the application lifecycle: application and infrastructure description models, deployment and reconfiguration, orchestration, and optimization. Members of the Magellan team contribute to four sub-topics including (1) Decentralized, market-based application orchestration for Fog and IoT environments; (2) Realizing and optimizing Serverless Computing in the Edge-Cloud continuum; (3) Orchestrating multi-dimensional resources in the Edge-Cloud continuum; and (4) Resource Provisioning for stream data processing in the Fog.

ARED TDFDE

Participants: Khaled Arsalane, Guillaume Pierre.

• Title:
  Traitement de données en flux distribué et à état
• Date/Duration:
  2022-2025
• Additionnal info/keywords:
  The PhD thesis of Khaled Arsalane is funded at 50% by the TDFDE project under the ARED program of Région Bretagne. This project investigates performance modeling and optimization of aggregation operators within geo-distributed data stream processing platforms.

ARED TANSIV

Participants: Léo Cosseron, Martin Quinson, Louis Rilling, Matthieu Simonin.

• Title:
  Time-Accurate Network Simulation Interconnecting VMs with Hardware Virtualization Towards Stealth Analysis
• Date/Duration:
  2022-2025
• Additionnal info/keywords:
  The PhD thesis of Léo Cosseron is funded at 50% by the ARED program “Breizh Cyber Valley” of Région Bretagne and 50% by the Creach Labs. This project studies how to defeat malware evasion techniques based on network performance analysis by adding a virtual realistic networks to analysis environments.

11.1.1 Scientific events: organisation

11.1.2 Scientific events: selection

11.1.3 Journal

11.1.4 Invited talks

• Anne-Cécile Orgerie gave a keynote at the EGC (Extraction et la Gestion des Connaissances) 2024 conference in Dijon, France.
• Guillaume Pierre gave an invited talk “Fog Computing Challenges in Natural Environment Observatories” at the Cloud Control Workshop on June 20th 2024 in Skåvsjöholm, Sweden.
• Martin Quinson gave an invited talk “SmolPhone : a smartphone with energy limits.” at the Eco-ICT winter school on October 11th 2024 in Beg Meil, France.

11.1.5 Leadership within the scientific community

• Anne-Cécile Orgerie is the director of the CNRS service group on ICT environmental impact (GDS EcoInfo).
• Anne-Cécile Orgerie is the chief scientist for the Rennes site of Grid'5000.

11.1.6 Scientific expertise

• Guillaume Pierre was a reviewer for a research proposal for the Austrian Science Fund (FWF).
• Anne-Cécile Orgerie was a reviewer for a research proposal for the Université Catholique de Louvain in Belgium.
• Anne-Cécile Orgerie was a reviewer for a research proposal for the Mitacs program in Canada.
• Anne-Cécile Orgerie was a member of the selection committees for an assistant professor position at University of Rennes in 2024.
• Cédric Tedeschi was a reviewer for two research proposals for the Natural Sciences and Engineering Research Council of Canada.
• Shadi Ibrahim was a member of the ACM Heidelberg Laureate Forum (HLF) 2024 Young Researcher Selection Committee.
• Shadi Ibrahim was a reviewer for a reserach proposal for the CEFIPRA Programme 2024: Indo-French Centre for the Promotion of Advanced Research.
• Shadi Ibrahim was a member of the jury (selection committee) for an assistant professor position at ENSAI Bruz in 2024.

11.1.7 Research administration

• Anne-Cécile Orgerie is an officer (chargée de mission) for the IRISA cross-cutting axis on Green IT.
• Anne-Cécile Orgerie is a member of the steering committee of the CNRS GDR RSD.

11.2.1 Teaching

• Bachelor: Marin Bertier, Programmation en Java, L1, INSA Rennes.
• Bachelor: Marin Bertier, Langage C, L3, INSA Rennes.
• Bachelor: Marin Bertier, Réseaux, L3, INSA Rennes.
• Bachelor: Shadi Ibrahim , Introduction to Big Data, ENSAI, Bruz.
• Bachelor: Francois Lemercier, Système Cloud et Réseaux, L3 Informatique, Univ. Rennes.
• Bachelor: Nikos Parlavantzas, Databases, L2, INSA Rennes.
• Bachelor: Guillaume Pierre, Systèmes Informatiques, L3 MIAGE, Univ. Rennes.
• Bachelor: Guillaume Pierre, Systèmes d'exploitation, L3 Informatique, Univ. Rennes.
• Bachelor: Martin Quinson , Architecture et Systèmes, L3 Informatique, ENS Rennes.
• Bachelor: Martin Quinson , Pédagogie, L3 Informatique, ENS Rennes.
• Master: Guillaume Pierre, Distributed Systems, M1 CR&CNI, Univ. Rennes.
• Master: Guillaume Pierre, Services for Cloud technology, M1 CR&CNI, Univ. Rennes.
• Master: Guillaume Pierre, Advanced Cloud Infrastructures, M2 CR&CNI&SIF, Univ. Rennes.
• Bachelor: Cédric Tedeschi, Systèmes, Cloud et Réseaux, L3 Informatique, Univ. Rennes.
• Master: Francois Lemercier, Network Administration and new technologies, M1 CNI, Univ. Rennes.
• Master: Francois Lemercier, Cloud, M1 RSSI, Univ. Rennes.
• Master: Nikos Parlavantzas, 4th-year Project, M1, INSA Rennes.
• Master: Nikos Parlavantzas, Clouds, M1, INSA Rennes.
• Master: Nikos Parlavantzas, IoT, M2, INSA Rennes.
• Master: Nikos Parlavantzas, Smart city services, M2 CNI, Univ. Rennes.
• Master: Nikos Parlavantzas, NoSQL, M2, Master for Smart Data Science, ENSAI.
• Master: Nikos Parlavantzas, SQL, M2, Master for Smart Data Science, ENSAI.
• Master: Martin Quinson , C++ system programming, ENS Rennes.
• Master: Martin Quinson , HPC programming, ENS Rennes.
• Master: Martin Quinson , Préparation à l'Agrégation d'Informatique (Networking, 20h ETD, Operating systems, 20h ETD), ENS Rennes.
• Master: Martin Quinson , Scientific Outreach, M2, 30 hEDT, ENS Rennes.
• Master: Cédric Tedeschi, Concurrences et Système d'Exploitation, M1 Informatique, Univ. Rennes.
• Master: Cédric Tedeschi, Concurrence et coopération dans les systèmes et les réseaux, M1 MIAGE. Univ. Rennes.
• Master: Cédric Tedeschi, Advanced Cloud Infrastructures, M2 Informatiquie, Univ. Rennes.
• Master: Cédric Tedeschi, Parallel Programming with Python, ENSAI
• Master: Marin Bertier, Systèmes d'exploitation, M1, INSA Rennes.
• Master: Shadi Ibrahim , Hadoop and Cloud Computing, 36hETD, M2 : Statistics for Smart Data, ENSAI, Bruz.
• Master: Shadi Ibrahim , NoSQL and Cloud technologies, 45hETD, M2, ENSAI, Bruz.
• Master: Anne-Cécile Orgerie , Energy efficiency in distributed systems, TelecomSudParis, Evry.

11.2.2 Supervision

• PhD defended: Chih-Kai Huang , “Scalability of Public Geo-Distributed Fog Computing Federations”, defended on December 9th 2024, supervised by Guillaume Pierre .
• PhD defended: Wedan Emmanuel Gnibga , “Modeling and optimizing edge computing infrastructures and their electrical system”, defended on November 19th 2024, supervised by Anne-Cécile Orgerie and Anne Blavette .
• PhD defended: Vladimir Ostapenco , “Modeling, evaluating and orchestrating heterogeneous leverages for large-scale cloud data centers management”, defended on December 18th 2024, supervised by Anne-Cécile Orgerie and Laurent Lefèvre .
• PhD in progress: Ammar Kazem , “Systèmes de traitement de données in-natura pour l'observation environnementale sous contrainte énergétique”, started in 2023, supervised by Guillaume Pierre and Laurent Longuevergne .
• PhD in progress: Khaled Arsalane , “Traitement de données en flux distribué et à état”, started in 2022, supervised by Guillaume Pierre .
• PhD in progress: Govind Kovilkkatt Panickerveetil , “Traitement de flux de données économe en énergie”, started in 2023, supervised by Guillaume Pierre and Romain Rouvoy .
• PhD in progress: Maxime Agusti , "Observation of baremetal co-location platforms, models and catalog proposal to reduce energy consumption", started in December 2021, supervised by Eddy Caron , Laurent Lefèvre and Anne-Cécile Orgerie .
• PhD in progress: Matthieu Silard , “Co-optimization of electrical and communication networks”, started in February 2023, supervised by Anne-Cécile Orgerie , Nicolas Montavont and Georgios Papadopoulos .
• PhD in progress: Matteo Chancerel , “Optimization of a fully distributed Fog powered by renewable energy sources”, started in September 2024, supervised by Anne-Cécile Orgerie .
• PhD in progress: Pablo Leboulanger , “Minimalist cloud, low on energy, hardware and software resources”, started in September 2024, supervised by Anne-Cécile Orgerie .
• PhD in progress: Thomas Stavis , “Replay of environmental leverages in Cloud infrastructures”, started in September 2024, supervised by Laurent Lefèvre and Anne-Cécile Orgerie .
• PhD in progress: Volodia Parol-Guarino , “Flexible resource allocation for FaaS applications in the fog”, started in October 2022, supervised by Nikos Parlavantzas
• PhD in progress: Mohamed Cherif Zouaoui Latreche ,“Balancing Performance and Sustainability for FaaS in the Fog”, started in October 2023, supervised by Nikos Parlavantzas and Hector Duran-Limon.
• PhD in progress: Mathis Valli , “Dynamic Adaptation of Machine Learning and Deep Learning Workflows across the Cloud-Fog-Edge Continuum“, started in April 2023, supvervised by Alexandru Costan, Loïc Cudennec and Cédric Tedeschi .
• PhD in progress: Mohammad Rizk , “Reliable and cost-efficient data placement and repair in P2P storage over immutable data“, started in November 2023, supervised by Shadi Ibrahim , Thomas Lambert , and Guillaume Pierre .
• PhD in progress: Quentin Acher , “Management of mutable data over P2P storage“, started in September 2023, supervised by Shadi Ibrahim and Claudia-Lavinia Ignat .
• PhD in progress: Marc Tranzer , “Energy efficient data management: Data reduction and protection meet performance and energy “, started in January 2024, supervised by Shadi Ibrahim and Guillaume Pierre .
• PhD in progress: Mathieu Laurent , “Efficient verification of asynchronous distributed programs”, started in October 2023, supervised by Martin Quinson and Thierry Jéron .
• PhD in progress: Leo Cosseron , “Time-Accurate Network Simulation Interconnecting VMs with Hardware Virtualization Towards Stealth Analysi”, started in October 2022, supervised by Martin Quinson and Louis Rilling .

11.2.3 Juries

• Guillaume Pierre was a reviewer of the PhD defense of Pierre-Antoine Rault (Université de Lorraine), December 12th 2024.
• Guillaume Pierre was chairperson at the PhD defense of Arthur Rauch (Université de Rennes), December 18th 2024.
• Anne-Cécile Orgerie was chairperson at the PhD defense of Vincent Rébiscoul (University of Rennes), May 14th 2024.
• Anne-Cécile Orgerie was chairperson at the PhD defense of Pierre Jacquet (University of Lille), July 19th 2024.
• Anne-Cécile Orgerie was chairperson at the PhD defense of Vincent Lannurien (ENSTA Bretagne), November 20th 2024.
• Anne-Cécile Orgerie was chairperson at the PhD defense of Chih-Kai Huang (University of Rennes), December 9th 2024.
• Nikos Parlavantzas was a reviewer for the Ph.D. defense of Armel Jeatsa Toulepi (Institut National Polytechnique de Toulouse), 11th September 2024.
• Cédric Tedeschi was the chairperson of the PhD defense of Thomas Bouvier (INSA de Rennes), November 4th, 2024.
• Cédric Tedeschi was the chairperson of the PhD defense of Timothé Albouy (Univ. Rennes), December 16th, 2024.
• Cédric Tedeschi was an examinor of the PhD defense of Manèle Ait Habouche (Université de Bretagne Occidentale), December 19th, 2024.
• Shadi Ibrahim was a thesis reader (member of the PhD Committee) of the PhD defense of Twinkle Jain (Northeastern University), USA, April 3rd 2024.

• Shadi Ibrahim co-organizies the SCI-Rennes seminar: a monthly series of scientific seminars for all staff at Inria research centre at Rennes University.

11.2.4 Productions (articles, videos, podcasts, serious games, ...)

• Anne-Cécile Orgerie was one of the authors of two publications in the popularization journal Bulletin 1024 of the Société Informatique de France in April 2024 28 and in November 2024 29.

11.2.5 Participation in Live events

• “L codent L créent” is an outreach program to send PhD students to teach Python to middle school students in 8 sessions of 45 minutes. Tassadit Bouadi (Lacodam) and Anne-Cécile Orgerie (Magellan) are coordinating the local version of this program, initiated in Lille. The first session in Rennes occured in April 2019, and a new session (the 6th) occured in 2024. The program is currently supported by: Fondation Blaise Pascal, ED MathSTIC, ENS de Rennes, Université de Rennes and Fondation Rennes 1.

11.2.6 Others science outreach relevant activities

• Guillaume Pierre gave an invited talk “Le cloud c'est bien, le fog c'est mieux!” to bachelor and master students of the University of Rennes on December 17th 2024.

International journals

• 4 articleS.Sylvain Bouveret, A.Aurélie Bugeau, A.-C.Anne-Cécile Orgerie and S.Sophie Quinton. De l'eau dans les nuages.Annales des Mines - Enjeux Numériques27September 2024, 40-47HALback to text
• 5 articleW. E.Wedan Emmanuel Gnibga, A.Anne Blavette and A.-C.Anne-Cécile Orgerie. Renewable Energy in Data Centers: the Dilemma of Electrical Grid Dependency and Autonomy Costs.IEEE Transactions on Sustainable Computing932024, 315-328HALDOIback to text
• 6 articleC.-K.Chih-Kai Huang and G.Guillaume Pierre. Aggregate Monitoring for Geo-Distributed Kubernetes Cluster Federations.IEEE Transactions on Cloud Computing2024, 1-15In press. HALDOIback to text

International peer-reviewed conferences

• 7 inproceedingsM.Maxime Agusti, E.Eddy Caron, B.Benjamin Fichel, L.Laurent Lefèvre, O.Olivier Nicol and A.-C.Anne-Cécile Orgerie. PowerHeat: A non-intrusive approach for estimating the power consumption of bare metal water-cooled servers.2024 IEEE International Conferences on Internet of Things (iThings) and IEEE Green Computing & Communications (GreenCom) and IEEE Cyber, Physical & Social Computing (CPSCom) and IEEE Smart Data (SmartData) and IEEE Congress on CybermaticsCopenhagen, Denmark2024, 1-7HALback to text
• 8 inproceedingsK.Khaled Arsalane, G.Guillaume Pierre and S.Shadi Ibrahim. Toward Stream Processing Elasticity in Realistic Geo-Distributed Environments.IC2E 2024 - 12th IEEE International Conference on Cloud EngineeringPaphos, Cyprus2024, 1-9HALback to text
• 9 inproceedingsL.Léo Cosseron, L.Louis Rilling, M.Matthieu Simonin and M.Martin Quinson. Simulating the Network Environment of Sandboxes to Hide Virtual Machine Introspection Pauses.EuroSec 2024 - 17th European Workshop on Systems SecurityAthène, GreeceApril 2024, 1-7HALDOIback to text
• 10 inproceedingsC.Clément Courageux-Sudan, A.-C.Anne-Cécile Orgerie and M.Martin Quinson. Studying the end-to-end performance, energy consumption and carbon footprint of fog applications.ISCC 2024 - 29th IEEE Symposium on Computers and CommunicationsParis, FranceIEEEJune 2024, 1-7HALback to text
• 11 inproceedingsW. E.Wedan Emmanuel Gnibga, A.Anne Blavette and A.-C.Anne-Cécile Orgerie. Energy-related Impact of Redefining Self-consumption for Distributed Edge Datacenters.IGSC 2024 - 15th International Green and Sustainable Computing ConferenceAustin, United States2024, 1-7HALback to text
• 12 inproceedingsW. E.Wedan Emmanuel Gnibga, A. A.Andrew A Chien, A.Anne Blavette and A.-C.Anne-Cécile Orgerie. FlexCoolDC: Datacenter Cooling Flexibility for Harmonizing Water, Energy, Carbon, and Cost Trade-offs.e-Energy 2024 - 15th ACM International Conference on Future and Sustainable Energy SystemsSingapore, SingaporeACMJune 2024, 108-122HALDOIback to text
• 13 inproceedingsC.-K.Chih-Kai Huang and G.Guillaume Pierre. UnBound: Multi-Tenancy Management in Scalable Fog Meta-Federations.UCC 2024 - 17th IEEE/ACM International Conference on Utility and Cloud ComputingSharjah, United Arab Emirates2024, 1-11HALback to text
• 14 inproceedingsC.Cherif Latreche, N.Nikos Parlavantzas and H. A.Hector A Duran-Limon. FoRLess: A Deep Reinforcement Learning-based approach for FaaS Placement in Fog.UCC 2024 - 17th IEEE/ACM International Conference on Utility and Cloud ComputingSharjah, United Arab Emirates2024, 1-9HALback to text
• 15 inproceedingsV.Vladimir Ostapenco, L.Laurent Lefèvre, A.-C.Anne-Cécile Orgerie and B.Benjamin Fichel. Exploring RAPL as a Power Capping Leverage for Power-Constrained Infrastructures.ICA3PP 2024 - 24th International Conference on Algorithms and Architectures for Parallel ProcessingMacau SAR, China2024, 1-10HALback to text
• 16 inproceedingsM.Matthieu Silard, A.-C.Anne-Cécile Orgerie, N.Nicolas Montavont and G.Georgios Papadopoulos. Local Peak Shaving for Electric Vehicles: A ready-to-deploy Smart Charging Solution.IGSC 2024 - 15th International Green and Sustainable Computing ConferenceAustin, United States2024, 1-7HALback to text
• 17 inproceedingsM.Mathis Valli, A.Alexandru Costan, C.Cédric Tedeschi and L.Loïc Cudennec. Towards Efficient Learning on the Computing Continuum: Advancing Dynamic Adaptation of Federated Learning.FlexScience'24: Proceedings of the 14th Workshop on AI and Scientific Computing at Scale using Flexible Computing InfrastructuresFlexScience 2024 - 14th Workshop on AI and Scientific Computing at Scale using Flexible Computing InfrastructuresPisa, ItalyACM2024, 42-49HALDOIback to text
• 18 inproceedingsH.Hao Wu, Y.Yue Yu, J.Junxiao Deng, S.Shadi Ibrahim, S.Song Wu, H.Hao Fan, Z.Ziyue Cheng and H.Hai Jin. StreamBox: A Lightweight GPU SandBox for Serverless Inference Workflow.ATC 2024 - USENIX Annual Technical ConferenceSanta Clara, United States2024HALback to text

National peer-reviewed Conferences

• 19 inproceedingsM.Matthieu Simonin and A.-C.Anne-Cécile Orgerie. Méthodologies de calculs de l'empreinte carbone sur une plateforme de calcul - L'exemple du site de Rennes de Grid'5000.JRES 2024 – Journée Réseaux de l'Enseignement SupérieurRennes, France2024, 1-13HALback to text

Edition (books, proceedings, special issue of a journal)

• 20 proceedingsSSDBM '24: Proceedings of the 36th International Conference on Scientific and Statistical Database Management.SSDBM 2024 - 36th International Conference on Scientific and Statistical Database ManagementACMAugust 2024, 1-165HALDOIback to text

Doctoral dissertations and habilitation theses

• 21 thesisC.-K.Chih-Kai Huang. Scalability of Public Geo-Distributed Fog Computing Federations.Université de RennesDecember 2024HALback to text

Reports & preprints

• 22 miscT.Thomas Lambert, S.Shadi Ibrahim and T.Twinkle Jain. On the (In)Accuracy of Straggler Detection in Big Data Analytic Systems.April 2024HALback to text

Other scientific publications

• 23 inproceedingsA.Ammar Kazem, G.Guillaume Pierre and L.Laurent Longuevergne. Enhancing Observatories With Fog Computing.SSDBM 2024 - 36th International Conference on Scientific and Statistical Database ManagementRennes, France2024, 1-1HALback to text
• 24 inproceedingsR.Radita Liem and S.Shadi Ibrahim. Towards Optimizing Data Movement in the Computing Continuum Workflow.SSDBM 2024 - 36th International Conference on Scientific and Statistical Database ManagementRennes, France2024HALback to text
• 25 inproceedingsA.Aloïs Rautureau, J.Joseph Paturel, M.Martin Quinson and S.Simon Rokicki. Quantifying the tiny-Small design of the SmolPhone.ICT4S 2024 - 10th International Conference on ICT for SustainabilityStockhlom, Sweden2024, 1-4HALback to text
• 26 inproceedingsM.Mohammad Rizk, S.Shadi Ibrahim and T.Thomas Lambert. Quantifying the Performance of Erasure Codes in P2P Storage Systems.SSDBM 2024 - 36th International Conference on Scientific and Statistical Database ManagementRennes, France2024HALback to text
• 27 inproceedingsM.Marc Tranzer and S.Shadi Ibrahim. On the Energy Footprint of Erasure Codes in Ceph.SSDBM 2024 - 36th International Conference on Scientific and Statistical Database ManagementRennes, France2024HALback to text

Scientific popularization

• 28 articleP.Philippe Ciblat, J.Jacques Combaz, M.Marceau Coupechoux, K.Kevin Marquet and A.-C.Anne-Cécile Orgerie. Effets environnementaux de la 5G (Partie 1) : La technologie 5G.1024 : Bulletin de la Société Informatique de France23April 2024, 53-72HALDOIback to text
• 29 articleP.Philippe Ciblat, J.Jacques Combaz, M.Marceau Coupechoux, K.Kevin Marquet and A.-C.Anne-Cécile Orgerie. Effets environnementaux de la 5G (partie 2) : Applications envisagées et acteurs impliqués.1024 : Bulletin de la Société Informatique de France24November 2024, 99-127HALDOIback to text