Proposition de stage 2009/2010
Topic: Benchmarking the performance of home gateways (or how fast is your DSL box?) Durée : 5 ou 6 mois
Équipe d’accueil :
Université Pierre et Marie Curie, Laboratoire LIP6-CNRS, équipe NPA (réseaux) Contact : renata.teixeira@lip6.fr
Responsables :
Renata Teixeira, Chargée de Recherche CNRS, LIP6 Fabian Schneider, Post-doc, LIP6
Mots clés :
Internet measurements, active and passive measurements, home gateways, characterization of path properties (reachability, delay, bandwidth)
Descriptif :
The NPA group at LIP6 participates in the “Nano Data Centers” project (NADA,
http://nanodatacenters.eu). The key idea of the NADA project is to take advantage of always-on DSL gateways or set-top-boxes for data hosting and delivery. This setting requires a new distributed (P2P- based) communication infrastructure that allows to access data dynamically, “on-demand”, and in real- time. At LIP6 we develop NADA’s monitoring infrastructure. Monitoring (e.g. determining delay or available bandwidth) is important to guarantee that the NADA platform is efficient (can reply to queries fast) and has high availability (users can always access content). Within NADA, monitoring will be deployed at the home gateways. Yet, the performance constraints of home gateways when used for monitoring are not well understood.
The goal of this internship is to explore and understand the performance limitations of off-the-shelf home gateways. As these low cost systems (DSL boxes) have significantly less resources as compared to laptops or workstations, we expect these systems to have limitations. These limitations might occur when forwarding network data from one interface to another or when running an analysis tool (this requires installing a new firmware) on the forwarded traffic. Reasons for that include small internal bandwidth, low processing power, and limited available memory. Therefore the task in this internship is to design a small test-bed and test cases. Then, the student will plug different DSL boxes into the test-bed to compare their performance. The insights of this work will be valuable for picking and designing the suitable devices for the Nano Data Centers project.
The challenges of this work are setting up a controlled environment to evaluate the performance of different equipment under different working conditions. Some experience with network measurement tools and scripting languages will be helpful. Besides knowledge in computer networks background in operating systems is appreciated. The student should develop scientific skills of experiment design, setup and conduction as well as scientific writing and presentation.
