ELAN

ELAN

Welcome to the Extended LAN (ELAN) project. ELAN is a project funded by ANR (Agence Nationale de la Recherche) for the period 2009-2011.

Partners

  • LIG - Laboratoire d'Informatique de Grenoble
  • Thales
  • Eurecom
  • IPANEMA

Executive Summary

The project concerns enterprise networks currently organized as private Intranets running over complex interconnections of layer 2 switches. Such networks commonly use Virtual LANs (VLANs) for traffic differentiation, Spanning Tree Protocol (STP) for reliability, and progressively introduce wireless APs for nomadic access. Home networks begin to follow a similar architecture, but on a small scale and with an increasing use of multimedia communications such as TV feeding. We can also find similar characteristics to enterprise Intranets in data centers that manage large numbers of blade servers interconnected through a structured topology of switches.

Traffic in these networks mainly involves the global Internet, but a significant part of communications stays local. In addition to bulk data flows either generated by user applications or by system tasks such as backup, real-time multimedia traffic of various form (VoIP, TV, video conferencing) becomes increasingly important, which raises the issue of an adequate support for QoS differentiation currently solved in a coarse grained manner by means of VLAN tagging and link overprovisioning. Moreover, enterprise and home networks become ever more complex with an increasing number of connected hosts and intermediate boxes to manage (NAT, filtering, wireless access points, application gateways etc.). At the same time, hosts benefit now from multiple network interfaces, e.g. standard laptops have at least one Ethernet and one 802.11 port, blade servers coming with at least two Ethernet ports.

The project proposes thus to address the issues of an improved interconnection at layer 2 (Extended LANs), scheduling techniques for supporting QoS, and seamless integration of wireless links in enterprise Intranets. We consider all these aspects in the light of ever evolving traffic, especially of multimedia nature. The project has three main stages. First, we propose to analyze current enterprise networks to understand the evolving nature of traffic and new requirements of applications. Then, we will design new techniques and tools for improving performance at the link layer, in particular we will address the issue of appropriate scheduling over wireless links. We will also propose new interconnection architectures that take into account the scale, multiple interfaces, and new requirements for QoS support. Finally, we will analyze how the designed tools and techniques will perform in environments supporting new types of traffic and evaluate their integration in chosen experiments.