The OneLab project.ppt

The OneLab project Thierry Parmentelat - INRIA A first (missed) attempt EU_planetlab was filed in spring within the SSA (Specific Support Action) framework not funded –ranked 2nd though too much focused on PL operations not innovative enough too much academic OneLab : a renewed attempt Within the STREP framework Specific Targeted REsearch Project Filed on September 21st 11-parties consortium Academic UPMC, INRIA, U’s : Madrid, Louvain, Napoli, Pisa Industrial Intel Research Cambridge, Ericsson Hungary, France Telecom, Alcatel Italia, Telekomunikacja Polska Decision expected in 2-3 weeks Orientation revisited More balanced between Operations (PLE) Innovation (add new features) Discussions between Experimental platforms vs Simulation And between Realistic vs Deterministic So you want to setup an experimental testbed Creating one’s local testbed Maximal flexibility Costly and time-consuming (admin user) Reaching real-scale requires agreements with –potentially many- other labs Tempting to elaborate on PlanetLab (partly) representative of the Internet, geographically scattered, wide impact in the community, still growing, cost-effective, very low-administration, unique management UI. What’s missing? PlanetLab still lacks a few features: poor support for wireless no/poor mobility control on the environment not deterministic (lacking framework for gathering measurements) Reproduce wireless links? Can be done for “stable” connectivity networks (UMTS/WiMax) Difficult for unstable connectivity (ad-hoc) Identify controllable aspects and define necessary extensions to attach such networks Or identify configurable parameters and run experiments with an “emulated” wireless network OneLab’s Objectives (2) Operations setup and run PLE a means to acquire autonomy would start with 20-30 hosts more if possible, no commitment newcomers welcome older nodes: how seamless is a move OneLab’s Objectives (3) New capabilities in nodes Native wireless support WiFi - ad hoc networ