Service discovery using a hypercube in PlanetLab

Dear all,

This is a first reminder that we are going to have a research meeting next Thursday 3rd at Campus Nord, Room C6-E220 at 12.00 am. Antonia will give us a talk about some interesting results from hers latest work, very related with her thesis.

 

 

 

 

Nowadays P2P solutions for large-scale distributed systems as Grid computing are the focus of multiple research endeavors. These solutions require service discovery in order to allocate services     efficiently. Even though many service discovery methods have been proposed, there have not been enough studies that focus on fault-tolerant search engines. In this work, a fault-tolerant service discovery solution for large-scale distributed systems is proposed and it is evaluated using PlanetLab. In the experiments performed, the failure rate was high, even in a supported and monitored network as PlanetLab. Therefore, it is necessary to address the phenomenon of failure variable, using adaptive protocols for delivering a satisfactory experience to user. The proposed search algorithm allows inter-domain service discovery with any number of nodes. The service discovery solution has been evaluated using a set of 150 nodes in PlanetLab.  The nodes are interconnected by using an overlay network with an incomplete hypercube topology. At the moment of a service request, the algorithm adapts to failures that inevitably occur in any large distributed system.