Large Scale Data Access: Architectures and Performance
Fabrizio Furano · ARCA (Università Ca' Foscari Venezia) · 2006
In this work we consider the problems related to the publishing of very large amounts of data to a big set of data consumers over a distributed environment. Different points of view of this problem have been considered, both methodological, algorithmic and technological. Some of the requirements for such systems are high availability, scalability, fast and low latency access to the data. Such a system has to be designed by starting considering the appropriate architectures, and then through reasoning about resource location algorithms and data transferring optimization techniques. The most important aspects of the performances of such a system are then analyzed through a simulation based approach, built around a case study: the publishing of data taken from the BaBar High Energy Physics experiment. The conclusions highlight the fact that the main objectives have been reached, and that such techniques have been successfully deployed.