Implementation of dataset staging process with improved security in a new analysis facility for ALICE experiment

초록

We run several computing facilities for scientific purpose based on Linux clusters. These facilities process many independent tasks simultaneously and require a large amount of data as input. A facility supports a research group in which they require on-the-fly random access to data remotely distributed on the Grid. To facilitate local storage rather than random access to the source, the facility should feature download and cache of remote data on user’s demand, which is called dataset staging. However, the facility was built several years ago and its operating system as well as data management system were outdated. In order to make the facility up-to-date, we needed to implement the dataset staging process based on the latest operating system as well as the data management system. Therefore we conducted thorough analysis on the behavior of the previous dataset staging process and its logs since the source code was not opened. In this paper, we describe the dataset staging process and discuss its implementation.

키워드