Team reso
Members
Overall Objectives
Scientific Foundations
Application Domains
Software
New Results
Contracts and Grants with Industry
Other Grants and Activities
Dissemination
Bibliography
previous up next PDF XML

Section: Application Domains

Panorama

RESO applies its research to the domains of high performance Cluster and Grid communications. Existing GRID applications did already identify potential networking bottlenecks, either caused by conceptual or implementation specific problems, or missing service capabilities. We participated to the elaboration of the first GGF document on this subject [88] [87] , [89] . Loss probability, important and incompressible latencies, dynamic behavior of network paths question profoundly models and technic used in parallel and distributed computing [81] . The particular challenge arises from a heavily distributed infrastructure with an ambitious end-to-end service demand. Provisioning end-to-end services with known and knowable characteristics in a large scale networking infrastructure requires a consistent service in an environment that spans multiple administrative and technological domains. The first bottleneck is often located at the interface between the local area network (LAN) and the wide area network (WAN). RESO conducted several actions in the field of Grid High Performance Networking in the context of the OGF, the European or National projects. These activities have been done in close collaboration with other INRIA and CNRS French teams (Grand Large, Mescal, Graal) involved in the GRID5000 and the Grid Explorer projects and other European teams involved in pfldnet and Glif communities. RESO joined the CARRIOCAS project which studies and implements a very high bit rate (up to 40 Gb/s per wavelength) network interconnecting super computers, storage systems and high resolution visualization device to support data and computing intensive applications in industrial and scientific domains. Our activities cover networking intelligence for high performance distributed applications.

Finally, the evolution of the Internet usage pushing the convergence of communication and computation at every level confirms our initial vision : the network should not be seen only as a black box providing pipes between edge machines, but as a vast cloud increasingly embedding the computational and storage resources to meet the requirement of emerging applications [6] . These resources are generally located at important crossroads and access points throughout the network. During the last few years we have seen that the distinction between packet forwarding and application processing has become blurred. The network community now starts to worry not only about forwarding packets without regard to application semantics, but is increasingly trying to exploit new functionalities within the network to meet the requirement of the application. Reciprocally, distributed systems and applications have traditionally been designed to run on top of the Internet, and to take the architecture of the Internet as given. The convergence of communication and computation at every level appears to be natural. It is however important to explore the full range of possibilities it can bring. Most of the proposals exploiting this convergence break the initial design philosophy of the Internet protocol stack (end to end argument for example), or if implemented in the application layer present lot of performance, resilience and scalability issues. We think that the Internet re-design raises the opportunity to better understand and assess higher-level system requirements, and use these as drivers of the lower layer architecture. In this process, mechanisms that are implemented today as part of applications may conceivably migrate into the network itself, and this is one of main driver of the researches of RESO and of our strong involvement in the new INRIA-BellLabs "Semantic Networking" research axis.


previous
next

Logo Inria