Tier 3 distributed data network

CoEPP user tasks

  • Analysis software development
  • Heavy I/O, non CPU intensive
  • * Data-reduction of large LHC datasets
  • CPU intensive
    • Simulation of physics events (if events saved large I/O)
    • Multi-dimensional function minimization for parameter estimation (optimization)

Typically data is accessed from compute nodes adjacent to storage e.g. Melbourne-based researcher reads and writes data from/to Melbourne storage using Melbourne compute nodes.

A distributed data network across Tier 3 sites has been deployed using xrootd, which is a file access protocol. The current configuration allows one to read CERN Root data files using the TFile::Open method. Queries for files are sent to redirectors hosted at sites, and which connect to data servers. A client queries the redirector for a file, the redirector checks whether its assigned data servers possess that file, and if so send the hostname and location of the file to the client.

Storage is distributed across four Australian cities, namely, Adelaide, Brisbane, Melbourne and Sydney. A diagram of the framework xrootd layer is given below.

tier3/faq/tier3_distributed_data_network.txt · Last modified: 2014/03/17 15:28 by antonio
Except where otherwise noted, content on this wiki is licensed under the following license: CC Attribution-Share Alike 4.0 International
Recent changes RSS feed Donate Powered by PHP Valid XHTML 1.0 Valid CSS Driven by DokuWiki