A5081007480- Caching and Content Delivery
- Cloud Computing and Resource Management
- Peer-to-Peer Network Technologies
- IoT and Edge/Fog Computing
- Distributed and Parallel Computing Systems
- Distributed systems and fault tolerance
- Advanced Data Storage Technologies
- Service-Oriented Architecture and Web Services
- Software System Performance and Reliability
- Opportunistic and Delay-Tolerant Networks
- Web Data Mining and Analysis
- Blockchain Technology Applications and Security
- Parallel Computing and Optimization Techniques
- IoT Networks and Protocols
- IPv6, Mobility, Handover, Networks, Security
- Wikis in Education and Collaboration
- Advanced Database Systems and Queries
- Network Traffic and Congestion Control
- Context-Aware Activity Recognition Systems
- Advanced Malware Detection Techniques
- Software-Defined Networks and 5G
- Data Stream Mining Techniques
- Wireless Networks and Protocols
- Data Management and Algorithms
- Green IT and Sustainability
Institut de Recherche en Informatique et Systèmes Aléatoires
2014-2024
Université de Rennes
2014-2024
Centre National de la Recherche Scientifique
2019-2024
Institut national de recherche en informatique et en automatique
1996-2024
Myriad (Germany)
2014-2023
American Jewish Committee
2023
University of Massachusetts Amherst
2023
Amazon (Germany)
2023
Karlstad University
2023
Hamburg University of Technology
2023
Improving resource provisioning of heterogeneous cloud infrastructures is an important research challenge. The wide diversity cloud-based applications and customers with different QoS requirements have recently exhibited the weaknesses current systems. Today's provide systems that dynamically adapt computational power by adding or releasing resources. Unfortunately, these scaling are fairly limited:(i) They restrict themselves to a single type resource, (ii)they unable fulfill in face spiky...
Developers often use replication and caching mechanisms to enhance Web application performance. The authors present a qualitative quantitative analysis of state-of-the art techniques used host applications. Their shows that selecting the best mechanism depends heavily on data workload requires careful review application's characteristics. They also propose technique for practitioners compare different mechanisms' performance their own
NoSQL cloud data stores provide scalability and high availability properties for web applications, but at the same time they sacrifice consistency. However, many applications cannot afford any inconsistency. CloudTPS is a scalable transaction manager which guarantees full ACID multi-item transactions issued by even in presence of server failures network partitions. We implement this approach on top two main families layers: Bigtable SimpleDB. Performance evaluation HBase (an open-source...
The transition from virtual machine-based infrastructures to container-based ones brings the promise of swift and efficient software deployment in large-scale computing infrastructures. However, fog environments which are often made very small computers such as Raspberry PIs, deploying even a simple Docker container may take multiple minutes. We demonstrate that makes inefficient usage available hardware resources, essentially using different subsystems (network bandwidth, CPU, disk I/O)...
The Distributed ASCI Supercomputer (DAS) is a homogeneous wide-area distributed system consisting of four cluster computers at different locations. DAS has been used for research on communication software, parallel languages and programming systems, schedulers, applications, applications. paper gives preview the most interesting results obtained so far in project.
We present Globule, a collaborative content delivery network developed by our research group. Globule is composed of Web servers that cooperate across wide area to provide performance and availability guarantees the sites they host. discuss issues involved in developing setting up large-scale CDN solutions for many its unique problems
To improve the scalability of Web, it is common practice to apply caching and replication techniques. Numerous strategies for placing maintaining multiple copies Web documents at several sites have been proposed. These approaches essentially a global strategy by which single family protocols used choose keep mutually consistent. We propose more flexible approach allowing each distributed document its own associated strategy. method assigning an optimal separately prove that generates...
Dynamic resource provisioning aims at maintaining the end-to-end response time of a web application within pre-defined SLA. Although topic has been well studied for monolithic applications, resources applications composed multiple services remains challenge. When SLA is violated, one must decide which service(s) should be reprovisioned optimal effect. We propose to assign an only front-end service. Other are not given any particular objectives. Services autonomously responsible their own...
ConPaaS is an open source platform-as-a-service environment that aims at simplifying the deployment of cloud applications. In ConPaaS, application defined as a composition one or more services. Each service self-managed and elastic: it can deploy itself on cloud, monitor its own performance, increase decrease processing capacity through dynamic resource provisioning.
Fog computing was designed to support the specific needs of latency-critical applications such as augmented reality, and IoT which produce massive volumes data that are impractical send faraway cloud centers for analysis. However this also created new opportunities a wider range in turn impose their own requirements on future fog platforms. This article presents study representative set 30 general-purpose platform should support.
Following the adoption of cloud computing, proliferation data centers in multiple regions, and emergence computing paradigms such as fog there is a need for integrated efficient management geo-distributed clusters. Geo-distributed deployments suffer from resource fragmentation, resources certain locations are over-allocated while others under-utilized. Orchestration platforms Kubernetes Federation offer conceptual models building blocks that can be used to build solutions address...
We present GlobeDB, a system for hosting Web applications that performs autonomic replication of application data. GlobeDB offers data-intensive the benefits low access latencies and reduced update traffic. The major distinction in our compared to existing edge computing infrastructures is process distribution data handled by automatically with very little manual administration. show significant performance gains can be obtained this way. Performance evaluations TPC-W benchmark over an...
This paper presents HotZone, an algorithm to place replicas in a wide-area network such that the client-to-replica latency is minimized. Similar previously proposed HotSpot algorithm, HotZone places on nodes along with their neighboring generate highest load. In contrast HotSpot, however, provides nearly-optimal results by considering overlapping neighborhoods. relies geometric model of Internet latencies, which effectively reduces cost placing K among N potential replica locations from...
Container technology has become a very popular choice for easing and managing the deployment of cloud applications services. orchestration systems such as Kubernetes can automate to large extent deployment, scaling, operations containers across clusters nodes, reducing human errors saving cost time. Designed with "traditional" environments in mind (i.e., datacenters close-by machines connected by high-speed networks), like present some limitations geo-distributed where computational...
Large swings in the demand for content are commonplace within Internet. When a traffic hotspot happens, however, there is delay before measures such as heavy replication of can be applied. This paper investigates potential predicting hotspots sufficiently far, albeit shortly, advance, so that preventive action taken hotpot takes place. Performing accurate load predictions appears to daunting challenge at first glance, but this shows that, when applied Web-server page-request traffic, even...
To enable decentralised actions in very large distributed systems, it is often important to provide the nodes with global knowledge about values of attributes across all nodes. This paper shows how, given an attribute whose are a system, each node can efficiently estimate statistical distribution these values. Simulations using heavily skewed real-world distributions show that our estimation methods outperform state-of-the-art heuristics by order magnitude average error 0.05% and maximum 2%....
Field-programmable gate arrays (FPGAs) can offer invaluable computational performance for many compute-intensive algorithms. However, to justify their purchase and administration costs it is necessary maximize resource utilization over expected lifetime. Making FPGAs available in a cloud environment would make them attractive new types of users applications help democratize this increasingly popular technology. there currently exists no satisfactory technique offering as resources sharing...
Container orchestration engines such as Kubernetes do not take into account the geographical location of application replicas when deciding which replica should handle request. This makes them ill-suited to act a general-purpose fog computing platforms where proximity between end users and serving is essential. We present proxy-mity, proximity-aware traffic routing system for distributed platforms. It seamlessly integrates in Kubernetes, provides very simple control mechanisms allow...