RESO

Optimized Protocols and Software for High-Performance Networks

  • Increase font size
  • Default font size
  • Decrease font size

This site is obsolete !!!
The RESO team is stopped since 31/12/2012.
RESO members have joined the Avalon and Dante teams

 

Optimized Protocols and Software

for High-Performance Networks

Advanced optical techniques allow core network infrastructures to rapidly improve their throughput, reliability, and flexibility. Links of 40 Gbit/s per wawelenght are now available and links of 100 Gbit/s are emerging. New technologies like 10 Gbit/s Ethernet or 10 Gbit/s Myrinet are also driving the increase of bandwidth in local area networks. These improvements have given the opportunity to create high-performance distributed systems that aggregate storage- and computing-resources into a virtual and integrated dynamic computing environment. Although these systems theoretically offer solutions for resource aggregation, a high and predictable performance for applications may be hard to obtain due to the unsuitability of the bandwidth-sharing paradigms (fairness, best effort, no QoS), communication protocols, and software overheads. In order to deliver this emerging traffic in a timely, cost-efficient, energy-efficient, and reliable manner over long-distance networks, several issues such as quality of service, security, traffic metrology, traffic modeling, and network-resource scheduling have to be investigated.

To address some of these issues, our work follows four major research topics :

  1. Protocols and Services Implementation
  2. Network Control and Optimization
  3. Network Metrology
  4. Network Services
inria-0058-0206.jpg

 

 

perspectives_final1.0.png

 

 

 

"The Future Internet will be a wide reservoir and a vast market for all kinds of small or huge resources and services for the communication and processing of information. Any entity, biological or not, will be able to access and combine those with the duration, trust-level, and capacity it will choose to pay for."
Pascale Vicat-Blanc Primet (march 2009)

 

RESO's methodology aims to combine theory, real experimentation, and software development. It is based on a three-step cycle:

  • fine analysis of the limits of classical approaches and existing protocols (TCP/IP)
  • proposition and exploration of alternative and disruptive solutions
  • theoretical and experimental evaluation of these proposals on large-scale facilities

This research focuses on a heavily ossified research object (the TCP/IP protocols of the Internet) and lies between a challenging emerging application domain on a specific network context. These factors induce a close interaction with both the application level and the underlying network level as well as a deep technical and scientifical knowledge of protocols and network equipments. The methodology is then based on a continous study of the high-end and original requirements and on experimental evaluations of the functionalities and performance of emeging, dedicated high-speed infrastructures. RESO gathers expertise in advanced high-performance local and cluster area networks protocols, in distributed systems and algorithmics, in protocol and protocol architecture design, in long-distance networking, in time-series analysis, and in statistical inference. This background work provides the basis for innovative protocols and software design. Moreover, we implement and experiment our proposals on real, emulated local or wide area testbeds with real conditions and large-scale applications.

 

Last Updated on Monday, 25 March 2013 19:06