6.1.1 Colony

• Name:
  Extending TCC+ and stronger guarantees to the network far edge
• Keywords:
  Edge Computing, Causal consistency, Replication and consistency, P2P, Hybrid consistency
• Functional Description:
  Colony is developed as an extension to AntidoteDB. It guarantees Transactional Causal Plus Consistency (TCC+) globally, based on a forest topology, rooted at the AntidoteDB data centres. It features P2P edge groups. An edge group enjoys strong consistency internally, functions with or without a cloud connection, and can disconnect and migrate to a different subtree in the topology. Colony features a dynamic access control system, leveraging the TCC+ guarantees between data and their associated access control lists.
• Contact:
  Ilyas Toumlilt
• Partner:
  Technische Universität Kaiserslautern (UniKL), Allemagne

Ordered Message broadcast. 

FIFO broadcast provides application ordering semantics of messages broadcast by the same sender and have been mostly implemented on top of unreliable static networks. In 18, we proposed a round-based FIFO broadcast algorithm with both termination detection and bounded message size for dynamic networks with recurrent connectivity. Initially, processes only know the number of processes N in the system and their identifier. Due to the dynamics of the network links, messages can be lost. Since no unbounded timestamp is used to identify a message, its size is bounded to $2N+O\left(log\right(N\left)\right)+msgSize$ bits where $msgSize$ is the bound size in bits of the broadcast data. We also proposed a FIFO atomic broadcast algorithm for dynamic networks with recurrent connectivity that uses the proposed FIFO broadcast and deliver primitives. This algorithm provides causal total order broadcast primitives.

Topology aware Leader election. 

Eventual leader election is an essential service for many reliable applications that require coordination actions on top of asynchronous fail-prone distributed systems.In 15, we propose CEL, a new distributed eventual leader election algorithm for dynamic networks, which exploits topological information to improve the choice of a central leader and reduce message exchanges. The algorithm has a crosslayer neighbors detection, with a neighbor-aware mechanism, to improve the sharing of topological knowledge and elect a central leader faster. It uses a self-pruning mechanism based on topological knowledge, combined with probabilistic gossip, to improve the performance of broadcast propagation. Evaluations were conducted on the OMNeT++ environment, simulating realistic MANET with interference, collision, and messages loss. The results show our approach reduces the number of messages sent and provides a stable leader and short paths to the leader.

Self-Stabilizing Leader election in Dynamic Networks. 

Essentially, self-stabilizing algorithms tolerate transient failures, since by definition such failures last a finite time (as opposed to crash failures, for example) and their frequency is low (as opposed to intermittent failures). Self-stabilization is also an approach to tolerate topological changes of the interconnection network. In that case, topological changes are considered as transient failures of links. So, as for other transient faults, the safety cannot be guaranteed all the time. Such approach becomes totally ineffective when the frequency of topological is very high. In this case, the network dynamics should be no more considered as an anomaly but rather as an integral part of the system nature.

We studied conditions under which stabilizing leader election can be solved in highly dynamic identified message passing systems. In 11, we provide necessary and sufficient conditions under which the problem can be solved assuming that every process can reach all the others at least once through a journey—a journey can be thought as a path over time from a source to a destination. In 10, we weaken the model: we consider dynamic systems where some processes may not be either sources or recipients. We show that self-stabilizing leader election can only be achieved if all processes are both sources and recipients—i.e., the models studied in  11. Otherwise (processes are either only sources or recipients), we show that even pseudo-stabilization—a weaker form of stabilization—cannot be achieved, except in a special case where each source can always reach all other processes within some bounded time.

Optimal Space Lower Bound for Deterministic Self-Stabilizing Leader Election Algorithms. 

Given a boolean predicate $\Pi$ on labeled networks (e.g., proper coloring, leader election, etc.), a self-stabilizing algorithm for $\Pi$ is a distributed algorithm that can start from any initial configuration of the network (i.e., every node has an arbitrary value assigned to each of its variables), and eventually converge to a configuration satisfying $\Pi$. It is known that leader election does not have a deterministic self-stabilizing algorithm using a constant-size register at each node, i.e., for some networks, some of their nodes must have registers whose sizes grow with the size $n$ of the networks. On the other hand, it is also known that leader election can be solved by a deterministic self-stabilizing algorithm using registers of $O(loglogn)$ bits per node in any $n$-node bounded-degree network. In 14, we show that this latter space complexity is optimal. Specifically, we prove that every deterministic self-stabilizing algorithm solving leader election must use $\Omega (loglogn)$-bit per node registers in some $n$-node networks. In addition, we show that our lower bounds go beyond leader election, and apply to all problems that cannot be solved by anonymous algorithms.

This last paper 14 received the Best Student Paper Award at OPODIS 2021.

Resource management in large networks. 

Network Operators expect to accurately satisfy a wide range of user's needs by providing fully customized services relying on Network Slicing. The efficiency of Network Slicing depends on an optimized management of network resources and Quality of Service (QoS). We focus on Network Slice placement optimization problem.

In 12 and 13, we consider online learning for optimal network slice placement under the assumption that slice requests arrive according to a non-stationary Poisson process. We propose a framework based on Deep Reinforcement Learning (DRL) combined with a heuristic to design algorithms. We specifically design two pure-DRL algorithms and two families of hybrid DRL-heuristic algorithms. To validate their performance, we perform extensive simulations in the context of a large-scale operator infrastructure. The evaluation results show that the proposed hybrid DRL-heuristic algorithms require three orders of magnitude of learning episodes less than pure-DRL to achieve convergence. This result indicates that the proposed hybrid DRLheuristic approach is more reliable than pure-DRL in a real non-stationary network scenario.

Adaptive stream processing systems. 

In 22, we propose a new adaptive Stream Processing System (SPS) for real-time processing that, based on input data rate variation, dynamically adapts the number of active operator replicas. Our SPS extends Storm by pre-allocating, for each operator, a set of inactive replicas which are activated (or deactivated) when necessary without the Storm reconfiguration cost. We exploit the MAPE model and define a new metric that aggregates the value of multiple metrics to dynamically changes the number of replicas of an operator. We deploy our SPS over Google Cloud Platform and results confirm that our metric can tolerate highly dynamic conditions, improving resource usage while preserving high throughput and low latency

Edge-first collaborative data systems. 

Distributing and replicating data at the edge enables immediate response, autonomy and availability in edge applications, such as gaming, cooperative engineering, or in-the-field information sharing. However, application developers and users demand the highest possible consistency guarantees, and specific support for group collaboration. To address this challenge, we designed the Colony system. It guarantees Transactional Causal Plus Consistency (TCC+) globally, dovetailing with Snapshot Isolation within edge groups. To help with scalability, fault tolerance and security, its logical communication topology is tree-like, with replicated roots in the core cloud, but with the flexibility to migrate a node or a group. Despite this hybrid approach, applications enjoy the same semantics everywhere in the topology. Our experiments show that peer groups improve throughput and response time significantly, performance is not affected in offline mode, and that migration is seamless.

This work, joint with Pierre Sutra of Télécom SudParis, is published at Middleware 2021 19 and is the thesis topic of Ilyas Toumlilt 34.

The concepts explored in this work are currently being industrialised, thanks to support from Inria Startup Studio, in the start-up Concordant.

Distributed and federated indexing

This work studies how to support efficient query processing, where users and data are distributed across multiple geographic locations. In particular, we study how the placement of indexes or materialized views, and the query/storage communication patterns, affect the metrics of performance, freshness, and resource consumption. We propose a query engine architecture that enables the administrator to to make the appropriate placement decisions on a case-by-case basis.

The enabling technique is a composition-based design. Our architecture consists of building block components, with a uniform interface and interaction semantics. Each block encapsulates some primitive query processing task. The administrator composes a system by instantiating pertinent blocks, connected in a directed acyclic graph. She chooses and places the blocks to provide some higher-order query processing capabilities, while satisfying the administrator's performance objectives.

We propose an implementation of the proposed approach, in the form of a framework for constructing and deployment application-specific query engines, called Proteus. The experimental evaluation supports the theoretical analysis of the trade-offs involved in query processing state placement, and suggests that Proteus can effectively occupy multiple different points in the design space of geo-distributed query processing.

This joint work with B. King, through a Cifre grant with Scality, is the thesis topic of D. Vasilas 35.

Exploring the coordination design space. 

Static analysis tools, for instance Soteria, make it possible to identify which updates conflict and need to be coordinated in order to maintain the data invariants. The coordination can be implemented in many ways, trading overhead against parallelism. The design space is multi-dimensional: locks can have various levels of granularity; different types of lock can be used (for example, mutex vs. shared/exclusive locks); the placement of the lock object has a significant impact. Furthermore, the performance of an option depends on the workload.

To systematize the dimensions of coordination, we construct a coordination lattice, which enables to systematically navigate the concurrency control dimensions of granularity, mode, and placement. The choice of granularity affects the cost of both lock acquisition and of lock contention (the coarser the lock, the lesser the cost of acquisition, but the higher the contention). Placement affects only the lock acquisition costs, and mode affects lock contention. Therefore the major dimension of the coordination lattice is granularity with mode and placement as secondary dimensions. Navigating any dimension has an impact on the overhead of locking. Accordingly, we propose a systematic approach to the design of correct coordination protocols, enabling the designer to select one according to performance metrics.

This work is published in the PhD thesis of Sreeja Nair 32.

A highly-available tree data type. 

The tree is an essential data structure in many applications. In a distributed application, such as a distributed file system, the tree is replicated. To improve performance and availability, different clients should be able to update their replicas concurrently and without coordination. Such concurrent updates converge if the effects commute, but nonetheless, concurrent moves can lead to incorrect states and even data loss. Such a severe issue cannot be ignored; ultimately, only one of the conflicting moves may be allowed to take effect. However, as such conflicts are infrequent, a solution should be lightweight. Previous approaches would require preventative cross-replica coordination, or totally order move operations after-the-fact, requiring roll-back and compensation operations.

In this work, we present a novel replicated tree that supports coordination-free concurrent atomic moves, and provably maintains the tree invariant. Our analysis identifies cases where concurrent moves are inherently safe, and we devise a lightweight, coordination-free, rollback-free algorithm for the remaining cases, such that a maximal safe subset of moves takes effect.

We present a detailed analysis of the concurrency issues with trees, justifying our replicated tree data structure. We provide mechanized proof that the data structure is convergent and maintains the tree invariant. Finally, we compare the response time and availability of our design against the literature.

This is joint work with Carla Ferreira of Universidade NOVA de Lisboa and her group. It is described in an Inria technical report 37 and in the PhD thesis of Sreeja Nair 32.

Highly-available file system. 

Building scalable and highly available geo-replicated file systems is hard. These systems need to resolve conflicts that emerge in concurrent operations in a way that maintains file system invariants, is meaningful to the user, and does not depart from the traditional file system interface. Conflict resolution in existing systems often leads to unexpected or inconsistent results. We design and implement ElmerFS, a geo-replicated, truly concurrent file system designed with the aim of addressing these challenges. ElmerFS is based on two key ideas: (1) the use of Conflict-Free Replicated Data Types (CRDTs) for representing file system structures, which ensures that replicas converge to a correct state, and (2) conflict resolution rules, which are determined by the choice of CRDT types and their composition, designed with the principle of being intuitive to the user.

This joint work with Romain Vaillant, Brad King and Dimitri Vasilas of Scality, was presented at HotStorage 2021 20.

A formally verified geo-distributed database, from model to implementation. 

We are leading an informal international collaboration to design and implement a full-featured geo-distributed database that is correct by construction. We adopt a stepwise approach. The initial step studies a simplistic concurrent database system; we describe its key invariants, formalise an operational semantics, and provide a reference implementation. From there, we plan to show that the semantic model satisfies the invariants (proof in Coq), that the implementation respects the model, and that the implementation passes litmus-test cases. Each following step adds a single feature, such as a cache, journaling, sharding, or garbage collection. We formalise the feature, and expect to show formally and experimentally that the improved system simulates the simpler one. We expect that the features compose, so that the correctness of the full system with all the features follows from the individual proofs. At the time of writing, the operational semantics for several steps is complete; the translation to Coq and the reference implementation are ongoing. This is joint work with Gustavo Petri (ARM Research Cambridge UK), Annette Bieniusa (TU Kaiserslautern) and Carla Ferreira (Universidade NOVA de Lisboa). A preliminary technical report is available 38. This is the PhD topic of Saalik Hatia.

Developing, deploying and running correct distributed software by composition. 

Modern applications are highly distributed and data-intensive. Programming a distributed system is challenging because of asynchrony, failures and trade-offs. In addition, application requirements vary with the use-case and throughout the development cycle. Moreover, existing tools come with restricted expressiveness or limited runtime customisability. This work aims to address this by improving reuse while maintaining fine-grain control and enhancing dependability. We argue that an environment for composable distributed computing will facilitate the process of developing distributed systems. We use high-level composable specification, verification tools and a distributed runtime.

This work will be the thesis topic of Benoît Martin and Laurent Prosperi (who is co-advised by Ahmed Bouajjani of Université de Paris).

Decentralised limiters for 5G Slicing. 

Meeting the goals of 5G networks —high bandwidth, low latency, massive connectivity, and resiliency— demands improvements to the infrastructure that hosts the network components. Mobile Network Operators will rely on a geographically distributed and highly scalable infrastructure that must handle and replicate user data consistently. This paper explores the management of user data with regards to data consistency in the first 5G specification. In particular we will focus on how the 5G system procedures handle and update data, and discuss failure scenarios where the correctness properties of the user data may be violated. In this work we present the necessary properties that an underlying data store must deliver in order to maintain correctness in the presence of failures.

The 5G specification describes a true geo-distributed system; we study the specification from the perspective of the consistency issues it raises. We currently focus on the specific use case of limiting resource usage in 5G slices (a slice is a virtual network involving geo-distributed users). We compare different consistency approaches and limitation algorithms, to study the trade-off between consistency, cost, and faithfulness to the prescribed limits.

This is joint work with Sofiane Imdali and Frédéric Martelli, through a Cifre grant with Orange Labs, and is the PhD topic of Jonathan Sid-Otmane 33.

In-kernel Caching

In 2021, we studied in-memory key-value stores aiming at improving their performances. In-memory key-value stores are critical components that help scale large internet services by providing low-latency access to popular data. Memcached, one of the most popular key-value stores, suffers from performance limitations inherent to the Linux networking stack and fails to achieve high performance when using high-speed network interfaces. While the Linux network stack can be bypassed using DPDK based solutions, such approaches require a complete redesign of the software stack and induce high CPU utilization even when client load is low. To overcome these limitations, we propose in 16, published at NSDI'21, BMC an in-kernel cache for Memcached that serves requests before the execution of the standard network stack. Requests to the BMC cache are processed as part of the NIC interrupts, which allows performance to scale with the number of cores serving the NIC queues. To ensure safety, BMC is implemented using eBPF. On small requests, our evaluations show that BMC improves throughput by up to 18x compared to the vanilla Memcached application and up to 6x compared to an optimized version of Memcached. In addition, our results also show that BMC has negligible overhead and does not decrease the throughput for larger requests.

9.1.1 STIC/MATH/CLIMAT AmSud project

9.2.1 Informal international cooperation

Let us report on the following informal international collaborations:

• Annette Bieniusa (TU Kaiserslautern, Germany) on the design and implementation of an efficient database backend. She co-advised the Masters' internship of Ayush Pandey.
• Annette Bieniusa (TU Kaiserslautern, Germany), Carla Ferreira (Universidade NOVA de Lisboa, Portugal) and Gustavo Petri (Université Paris-Diderot, then ARM Research, Cambridge, UK), on the formalisation and proof of an advanced distributed database. This is part of the thesis topic of Saalik Hatia.
• Annette Bieniusa (TU Kaiserslautern, Germany), Nuno Preguiça and João Leitão (Universidade NOVA de Lisboa, Portugal), and Carlos Baquero (Universidade do Minho, Portugal), on distributed data management and consistency. This is relevant to several research projects in the group.
• Carla Ferreira (Universidade NOVA de Lisboa, Portugal) on the design and proof of a highly-available replicated tree data structure. This work is reported in the PhD thesis of Sreeja Nair 32.

9.2.2 Visits of international scientists

9.3.1 ANR

9.3.2 Informal national cooperation

Let us report on some additional national collaborations:

• Ahmed Bouajjani (Université de Paris), on the design of a programming environment for building distributed systems by correct composition. Ahmed Bouajjani is PhD co-advisor of Laurent Prosperi.
• Pierre Sutra (Télécom SudParis, France), on consistency protocols, especially for edge databases. Pierre Sutra contributed to the Colony project 19.

10.1.1 Scientific events: organisation

10.1.2 Scientific events: selection

10.1.3 Journal

10.2.1 Research administration

• Colette Johnen, since 2020: Member of section 27 of Conseil national des Universités.
• Pierre Sens, until Aug 2021: Member of Section 6 of the national committee for scientific research CoNRS.
• Franck Petit, Pierre Sens, since 2012: Member of the Executive Committee of Labex SMART, CoChairs of Track 4, Autonomic Distributed Environments for Mobility.

10.3.1 Teaching

• Julien Sopena is Member of “Directoire des formations et de l'insertion professionnelle” of Sorbonne Université, France
• Master: Julien Sopena is responsible of Computer Science Master's degree in Distributed systems and applications (in French, SAR), Sorbonne Universités, France
• Master: Luciana Arantes, Swan Dubois, Jonathan Lejeune, Franck Petit, Pierre Sens, Julien Sopena, Advanced distributed algorithms, M2, Sorbonne Université, France
• Master: Jonathan Lejeune, Designing Large-Scale Distributed Applications, M2, Sorbonne Université, France
• Master: Maxime Lorrillere, Julien Sopena, Linux Kernel Programming, M1, Sorbonne Université, France
• Master: Luciana Arantes, Swan Dubois, Jonathan Lejeune, Pierre Sens, Julien Sopena, Operating systems kernel, M1, Sorbonne Université, France
• Master: Luciana Arantes, Swan Dubois, Franck Petit, Distributed Algorithms, M1, Sorbonne Université, France
• Master: Franck Petit, Autonomic Networks, M2, Sorbonne Université, France
• Master: Franck Petit, Distributed Algorithms for Networks, M1, Sorbonne Université, France
• Master: Jonathan Lejeune, Julien Sopena, Client-server distributed systems, M1, Sorbonne Université, France.
• Master: Luciana Arantes, Pierre Sens, Franck Petit. Cloud Computing, M1, EIT Digital Master, France.
• Master: Julien Sopena, Marc Shapiro, Ilyas Toumlilt, Francis Laniel. Kernels and virtual machines (Noyaux et machines virtuelles, NMV), M2, Sorbonne Université, France.
• Licence: Pierre Sens, Luciana Arantes, Julien Sopena, Principles of operating systems, L3, Sorbonne Université, France
• Licence: Swan Dubois, Initiation to operating systems, L3, Sorbonne Université, France
• Licence: Swan Dubois, Multi-threaded Programming, L3, Sorbonne Université, France
• Licence: Jonathan Lejeune, Oriented-Object Programming, L3, Sorbonne Université, France
• Licence: Franck Petit, Advanced C Programming, L2, Sorbonne Université, France
• Licence: Swan Dubois, Jonathan Lejeune, Franck Petit, Julien Sopena, Introduction to operating systems, L2, Sorbonne Université, France
• Licence: Mesaac Makpangou, C Programming Language, 27 h, L2, Sorbonne Université, France
• Ingénieur 4ème année : Marc Shapiro, Introduction aux systèmes d'exploitation, 26 h, M1, Polytech Sorbonne Université, France.
• Licence : Philippe Darche (coordinator), Architecture of Internet of Things (IoT), 2 × 32h, L3, Institut Universitaire Technologique (IUT) Paris Descartes, France.
• Engineering School: Philippe Darche (coordinator), Solid-State Memories, 4th year, ESIEE, France.
• DUT: Philippe Darche (coordinator), Introduction to Computer Systems - Data representation, 60h, Institut Universitaire Technologique (IUT) Paris Descartes, France.
• DUT: Philippe Darche (coordinator), Computer Architecture, 32h, Institut Universitaire Technologique (IUT) Paris Descartes, France.
• DUT: Philippe Darche (coordinator), Computer Systems Programming, 80h, Institut Universitaire Technologique (IUT) Paris Descartes, France.

10.3.2 PhD Advising

• CIFRE PhD: José Alves Esteves, "Adaptation dynamique en environnements répartis contraints", Sorbonne Univ. Dec. 2021. Advised by Pierre Sens and Amina Boubendir Orange Labs.
• PhD: Sreeja Nair, “Designing safe and highly available distributed applications,” Sorbonne Univ., Jul. 2021. Advised by Marc Shapiro.
• CIFRE PhD: Jonathan Sid-Otmane. “Étude des contraintes de cohérence des données dans la 5G, appliquée aux limitations d'usage de ressources dans les slices réseau,” Dec. 2021. Advised by Marc Shapiro with Sofiane Imadali and Frédéric Martelli, Orange Labs.
• PhD: Ilyas Toumlilt, “Colony: A Hybrid Consistency System for Highly-Available Collaborative Edge Computing,” Sorbonne Univ., Dec. 2021. Advised by Marc Shapiro.
• CIFRE PhD: Dimitrios Vasilas, “A flexible and decentralised approach to query processing for geo-distributed data systems”, Sorbonne Univ., Jul. 2021. Advised by Marc Shapiro, with Brad King, Scality.
• PhD in progress: Aymeric Agon-Rambosson, "Maintien du groupes dans un environnement hautement hétérogène et dynamique", Sorbonne Univ., since Oct. 2021. Advised by Pierre Sens and Jonathan Lejeune.
• PhD in progress: Arnaud Favier, "Election de leader dans les réseaux dynamiques", Sorbonne Univ., since Sep. 2018. Advised by Pierre Sens and Luciana Arantes.
• PhD in progress: Célia Mahamdi, "Prise de décision collaborative dans un système distribué et dynamique", Sorbonne Univ., since Sep. 2020. Advised by Mesaac Makpongou and Jonathan Lejeune.
• PhD in progress: Saalik Hatia, “Efficient management of memory and storage for CRDTs,” Sorbonne Univ., since Oct. 2018. Advised by Marc Shapiro.
• PhD in progress: Gabriel Le Bouder, "Autonomic synchronization", Sorbonne Univ., since Sep. 2019. Advised by Franck Petit.
• PhD in progress: Benoît Martin, “Protocol de cohérence hybride: de la cohérence causale à la cohérence forte,” Sorbonne Univ., since Sep. 2019. Advised by Mesaac Makpangou and Marc Shapiro.
• PhD in progress: Laurent Prosperi, “Abstractions, langage et runtime pour les systèmes distribués,” Sorbonne Univ., since Sep. 2019. Advised by Marc Shapiro.
• PhD in progress: Daniel Wladdimiro, "Adaptation dynamique en environnements répartis contraints", Sorbonne Univ., since Sep. 2019. Advised by Pierre Sens and Luciana Arantes.
• PhD in progress: Daniel Wilhelm, "Algorithmes de diffusion causale dans les systèmes répartis dynamique", Sorbonne Univ., since Oct. 2019, Pierre Sens and Luciana Arantes.

10.3.3 Juries

Pierre Sens was the reviewer of

• Flavien Vernier, HDR, LISTIC, Univ. Savoie Mont Blanc
• Mozhdeh Farhadi, PhD, IRISA, Univ. Rennes 1
• Pedro Penna, PhD, LIG, Univ. Grenoble Alpes
• Ugaitz Amozarrain, PhD, San Sebastien, Univ. Basque (Espagne)

Pierre Sens was Chair of

• Tuanir França Rezende, PhD, SAMOVAR, Inst. Polytechnique Paris
• Jonathan Sid-Otmane, PhD, LIP6, Sorbonne Univ.

International journals

• 9 articleS.Stéphane Devismes, A.Anissa Lamani, F.Franck Petit, P.Pascal Raymond and S.Sébastien Tixeuil. Terminating Exploration Of A Grid By An Optimal Number Of Asynchronous Oblivious Robots.The Computer Journal641January 2021, 132-154
  HALDOI

International peer-reviewed conferences

• 10 inproceedingsK.Karine Altisen, S.Stéphane Devismes, A.Anaïs Durand, C.Colette Johnen and F.Franck Petit. On Implementing Stabilizing Leader Election with Weak Assumptions on Network Dynamics.PODC '21: ACM Symposium on Principles of Distributed ComputingVirtual Event, ItalyACMJuly 2021, 21-31
  HALDOIback to text
• 11 inproceedingsK.Karine Altisen, S.Stéphane Devismes, A.Anaïs Durand, C.Colette Johnen and F.Franck Petit. Self-stabilizing Systems in Spite of High Dynamics.22nd International Conference on Distributed Computing and Networking, ICDCN'21ICDCN '21: International Conference on Distributed Computing and Networking 2021Nara, JapanLaBRI, CNRS UMR 5800January 2021, 156–165
  HALDOIback to textback to text
• 12 inproceedingsJ. J.Jose Jurandir Alves Esteves, A.Amina Boubendir, F.Fabrice Guillemin and P.Pierre Sens. DRL-based Slice Placement Under Non-Stationary Conditions.CNSM 2021 - 17th International Conference on Network and Service ManagementIzmir, TurkeyOctober 2021
  HALback to text
• 13 inproceedingsBest paperJ. J.Jose Jurandir Alves Esteves, A.Amina Boubendir, F.Fabice Guillemin and P.Pierre Sens. DRL-based Slice Placement under Realistic Network Load Conditions.CNSM 2021 - 17th International Conference on Network and Service ManagementIzmir, TurkeyOctober 2021
  HALback to textback to text
• 14 inproceedingsBest paperL.Lélia Blin, L.Laurent Feuilloley and G.Gabriel Le Bouder. Optimal Space Lower Bound for Deterministic Self-Stabilizing Leader Election Algorithms.OPODISOPODIS 2021 - International Conference on Principles of Distributed SystemsLIPIcsStrasbourg, FranceDecember 2021
  HALback to textback to textback to text
• 15 inproceedingsBest paperA.Arnaud Favier, L.Luciana Arantes, J.Jonathan Lejeune and P.Pierre Sens. Centrality-Based Eventual Leader Election in Dynamic Networks.NCA 2021C- 20th IEEE International Symposium on Network Computing and ApplicationsCambridge, Boston, United StatesNovember 2021
  HALback to textback to text
• 16 inproceedingsY.Yoann Ghigoff, J.Julien Sopena, K.Kahina Lazri, A.Antoine Blin and G.Gilles Muller. BMC: Accelerating Memcached using Safe In-kernel Caching and Pre-stack Processing.NSDI'21 - 18th USENIX Symposium on Networked Systems Design and ImplementationVirtual event, United StatesUSENIX AssociationApril 2021, 487-501
  HALback to text
• 17 inproceedingsH.Hasan Heydari, G.Guthemberg Silvestre and L.Luciana Arantes. Efficient Consensus-Free Weight Reassignment for Atomic Storage.NCA 2021 - 20th IEEE International Symposium on Network Computing and ApplicationsVirtual, FranceNovember 2021
  HAL
• 18 inproceedingsC.Colette Johnen, L.Luciana Arantes and P.Pierre Sens. FIFO and Atomic broadcast algorithms with bounded message size for dynamic systems.SRDS 2021 - 40th International Symposium on Reliable Distributed SystemsChicago / Virtual, United StatesSeptember 2021
  HALback to text
• 19 inproceedingsI.Ilyas Toumlilt, P.Pierre Sutra and M.Marc Shapiro. Highly-available and consistent group collaboration at the edge with colony.Middleware '21: Proceedings of the 22nd International Middleware ConferenceMiddleware 2021: 22nd International Middleware ConferenceQuébec / Virtual, CanadaACMDecember 2021, 336–351
  HALDOIback to textback to text
• 20 inproceedingsR.Romain Vaillant, D.Dimitrios Vasilas, M.Marc Shapiro and T. L.Thuy Linh Nguyen. CRDTs for truly concurrent file systems.HotStorage '21 -13th ACM Workshop on Hot Topics in Storage and File SystemsVirtual, FranceJuly 2021
  HALback to text
• 21 inproceedingsD.Daniel Wilhelm, L.Luciana Arantes and P.Pierre Sens. A scalable causal broadcast that tolerates dynamics of mobile networks.23rd International Conference on Distributed Computing and Networking (ICDCN)New Delhi / Virtual, IndiaJanuary 2022
  HAL
• 22 inproceedingsD.Daniel Wladdimiro, L.Luciana Arantes, P.Pierre Sens and N.Nicolas Hidalgo. A Multi-Metric Adaptive Stream Processing System.NCA 2021 - 20th IEEE International Symposium on Network Computing and ApplicationsCambridge, Boston, United StatesNovember 2021
  HALback to text

Conferences without proceedings

• 23 inproceedingsN.Nathan Maurice, J.Julien Sopena and L.Lionel Lacassagne. Un nouvel algorithme efficace de Split & Merge pour systèmes embarqués.COMPAS 2021 - Conférence francophone d'informatique en Parallélisme, Architecture et SystèmeLyon, FranceJuly 2021
  HAL

Scientific books

• 24 bookP.Philippe Darche. Le Microprocesseur 1 : fonctions de calcul et de mémorisation, modèles de calcul et architecture des ordinateurs.ISTE LtdJune 2021
  HAL
• 25 bookP.Philippe Darche. Le Microprocesseur 2 : communication dans un système numérique.ISTE LtdJune 2021
  HAL
• 26 bookP.Philippe Darche. Le Microprocesseur 3 : aspects matériels.ISTE LtdJune 2021
  HAL
• 27 bookP.Philippe Darche. Le Microprocesseur 4 : aspects logiciels.ISTE LtdJune 2021
  HAL
• 28 bookP.Philippe Darche. Le Microprocesseur 5 : aspects logiciels et matériels du développement, du débogage et du test.ISTE LtdJune 2021
  HAL
• 29 bookP.Philippe Darche. Microprocessor 4. Core Concepts: Software Aspects.ISTE Ltd and John Wiley & Sons, Inc.February 2021
  HAL
• 30 bookP.Philippe Darche. Microprocessor 5. Software and Hardware Aspects of Development, Debugging and Testing – The Microcomputer.ISTE Ltd and John Wiley & Sons, Inc.February 2021
  HAL

Doctoral dissertations and habilitation theses

• 31 thesisJ. J.Jose Jurandir Alves Esteves. Optimization of Network Slice Placement in Distributed Large Scale Infrastructures From Heuristics to Controlled Deep Reinforcement Learning.Sorbonne Universites, UPMC University of Paris 6; Orange LabsDecember 2021
  HAL
• 32 thesisS. S.Sreeja Sasidharan Nair. Designing safe and highly available distributed applications.Sorbonne UniversitéJuly 2021
  HALback to textback to textback to text
• 33 thesisJ.Jonathan Sid-Otmane. A study of data consistency constraints in 5G, applied to limiting resource usage in network slices.Sorbonne UniversiteDecember 2021
  HALback to text
• 34 thesisI.Ilyas Toumlilt. Colony: A Hybrid Consistency System for Highly-Available Collaborative Edge Computing.Sorbonne UniversitéDecember 2021
  HALback to text
• 35 thesisD.Dimitrios Vasilas. A flexible and decentralised approach to query processing for geo-distributed data systems.Sorbonne UniversitéFebruary 2021
  HALback to text

Reports & preprints

• 36 reportC.Colette Johnen and M.Mohammed Haddad. Efficient self-stabilizing construction of disjoint MDSs in distance-2 model.Inria Paris, Sorbonne Université; LaBRI, CNRS UMR 5800; LIRIS UMR CNRS 5205February 2021
  HAL
• 37 reportS. S.Sreeja S Nair, F.Filipe Meirim, M.Mário Pereira, C.Carla Ferreira and M.Marc Shapiro. A coordination-free, convergent, and safe replicated tree.RR-9395LIP6, Sorbonne Université, Inria de Paris; Universidade nova de LisboaFebruary 2021, 36
  HALback to text