Grid'5000 user report for Laurent Lefevre

Jump to: navigation, search

User information

Laurent Lefevre (users, user, account-manager, site-manager, lyon, g5kschool, ml-users user)
More user information in the user management interface.

Experiments

  • LSCAN (Large Scale Autonomic Networks) (Networking) [achieved]
    Description: Large Scale Autonomic Networks : we want to experiment a networking infrastructure based on active and programmable network equipments. Execution environments deployed are based on Tamanoir software suite developped by Gelas&Lefevre (INRIA RESO). Active Networks, allowing users or applications to inject customized programs into the nodes of the network, are considered as a promising way to enhance the common idea of what a network can do. The creation of new services, such as on-the-fly compression or encryption, and what is more important a new way to think about development and deployment of customized modules to perform computation within the network can lead to massive improvements to network functionality. The Tamanoir project, based on active networks where a new generation of network equipments (such as routers, proxies, and gateways) apply services on the fly on data packets, deals with, in a successful way, some of the historical problems that this kind of approach has to face, such as security and high performance, implementing an efficient multi-streams active transport and dynamic services deployment in the net. Tamanoir services may operate either at a lower level (packets marking, selective drops...) or at a higher one (QoS, cryptography, compression on the fly...) The Tamanoir architecture design does not interphere with the core network, mainly to guarantee higher performance results, and it's deployed only on the network periphery. Tamanoir Active Nodes (TAN) provide persistent active nodes which are able to handle different applications and various data stream at the same time. The two main transport protocol (TCP and UDP) are supported by the TAN for carrying data. We use the ANEP (Active Network Encapsulated Protocol) format to send data over active networks. The injection of new functionalities, called services, is independent from the data stream: services are deployed on demand when streams reach an active node which does not hold the required service. There are two ways for service deployment: with a service repository, where TANs send all requests for downloading required services, and without, in which case the TAN queries the active node that sent the stream for the service. When the service is installed in memory, it is ready to process the stream. It is worth noticing that a stream can cross equally a classical router, obviously, without any processing actions.
    Results: In the context of the Grid5000 project, we have developped the LSCAN software suite which is dedicated to the deployment (graphical), management (trough web services) and experiment of large scale deployment of autonomic network nodes. LSCAN is adapted from the Nagios software suite and currently validated on Grid5000 platform.
    illustrating chart picture not found
    More information here
  • Large scale evaluation of router assisted transport protocols : the XCP-i approach (Networking) [achieved]
    Description: XCP (eXplicit Control Protocol) is a transport protocol that uses the assistance of specialized routers to very accurately determine the available bandwidth along the path from the source to the destination. In this way, XCP efficiently controls the sender's congestion window size thus avoiding the traditional slow-start and congestion avoidance phase. However, XCP requires the collaboration of all the routers on the data path which is almost impossible to achieve in an incremental deployment scenario of XCP. It has been shown that XCP behaves badly, worse than TCP, in the presence of non-XCP routers thus limiting dramatically the benefit of having XCP running in some parts of the network. In this work, we address this problem and propose XCP-i which is operable on an internetwork consisting of XCP routers and traditional IP routers without loosing the benefit of the XCP control laws. XCP-i basically executes the next four steps to discover and compute a new feedback that reflects the state of the network where non-XCP routers are placed: - Discover where the non-XCP routers are in the data path. - Discover the upstream and downstream XCP-i routers of the non-XCP routers. - Estimate the available bandwidth where the non-XCP routers are placed. - Create a virtual XCP-i router that computes a new feedback using the estimated available bandwidth before. The simulation results on a number of topologies that reflect the various scenario of incremental deployment on the Internet show that although XCP-i performances depend on available bandwidth estimation accuracy, XCP-i still outperforms TCP on high-speed links. On the Grid5000 platform, we want to evaluate XCP-i approach on a large scale. An XCP-i software based on a linux kernel is currently under development.
    Results:
    illustrating chart picture not found
    More information here
  • Monitoring Grid resources usage with SBLOMARS (Middleware) [achieved]
    Description: We want to monitor Grid resources usage by deploying SBLOMARS agents collecting usage data. Experiments done with the collaboration of Edgar Magana (UPC Barcelona, Spain).
    Results:
    illustrating chart picture not found
    More information here
  • Large Scale DSM systems (Middleware) [achieved]
    Description: The VODCA system will be experimented on large scale infrastructure (more than few hundreds of nodes). Collaboration work with University of Otago, New Zealand.
    Results:
    More information here
  • Towards an Energy Efficient Grid5000 (Networking) [in progress]
    Description: This project explores the design of energy-aware software frameworks dedicated to large scale distributed systems. These frameworks will collect energy usage information and provide them to resources managrs and schedulers. Large scale experimental validations on Grid5000 and DSLLAB platforms will be proposed.

    This research activity is part of the ARC Green-Net project supported by INRIA.

    Live energy monitoring information can be obtained here .

    Participants of this experiment from INRIA RESO Project Team :
    • Laurent Lefevre
    • Anne Cecile Orgerie
    • Jean-Patrick Gelas

    Results:
    illustrating chart picture not found
    More information here
  • Autonomic Internet Projet (Networking) [achieved]
    Description: On-demand deployment of services on large scale virtual network infrastructures on Grid'5000. Design and development of the Autonomic Network Programming Interface (ANPI) and OVNI5000. This research is part of the European STREP : Autonomic Internet Project.
    Results: The project Autoi has been deployed on Grid5000. A new framework/software for the manage and deploy has been created : OVNI5000.
    More information here

Publications

Collaborations

    Success stories and benefits from Grid'5000

    • Overall benefits
    • The Grid5000 platform will allow us to deploy a large reproductible environment for experimenting interactions between programmable network equipments and large number of clients. This large scale experiment permits to evaluate scalability of the overall approach in terms of number of streams, clients and equipments.

    last update: 2011-10-20 16:12:55

    Personal tools
    Namespaces

    Variants
    Views
    Actions
    Public Portal
    Users Portal
    Admin portal
    Wiki special pages
    Toolbox