14–17 May 2013
The University of Tokyo
Asia/Tokyo timezone

Session

Networks in Production and Analysis Workflows

15 May 2013, 10:00
The University of Tokyo

The University of Tokyo

Sanjo-Kaikan Hall, Hongo Campus 7-3-1, Hongo, Bunkyo-ku, Tokyo https://www.icepp.s.u-tokyo.ac.jp/index-e.html

Conveners

Networks in Production and Analysis Workflows

  • Stephane Jezequel (Centre National de la Recherche Scientifique (FR))
  • Michael Ernst (BNL)
  • Simone Campana (CERN)

Networks in Production and Analysis Workflows

  • Michael Ernst (Unknown)
  • Stephane Jezequel (Centre National de la Recherche Scientifique (FR))
  • Simone Campana (CERN)

Presentation materials

There are no materials yet.

  1. Simone Campana (CERN)
    15/05/2013, 10:00
  2. Alessandro Di Girolamo (CERN), Cedric Serfon (CERN)
    15/05/2013, 10:15
    - Introduction, status and deployment models - How will Rucio use and interact with FTS 3? - What is the right level for "network-awareness" - the transfer scheduler (FTS) or the DDM layer?
    Go to contribution page
  3. Shawn Mc Kee (University of Michigan (US))
    15/05/2013, 10:45
    Where are we regarding deployment, functionality, benefit and what is missing?
    Go to contribution page
  4. Ilija Vukotic (University of Chicago (US)), Wahid Bhimji (University of Edinburgh (GB)), Wei Yang (SLAC National Accelerator Laboratory (US))
    15/05/2013, 11:10
  5. Michael Ernst (Unknown)
    15/05/2013, 13:00
    We can dynamically acquire 1000's of processors from commercial clouds, but can we also get adequate network connectivity, just in time we need it?
    Go to contribution page
  6. Hironori Ito (BNL)
    15/05/2013, 13:20
    Are our transfer engines delivering optimal per-stream performance, what is limiting the performance?
    Go to contribution page
  7. Doug Benjamin (Duke University (US))
    15/05/2013, 13:45
    - Data movement in end-to-end analysis work flows - Implications of the new analysis model proposed by the Analysis Model Study Group - How much data do we replicate through PD2P (for long-term usage) vs. planned replication? What would be the right mix in the future given limited disk space?
    Go to contribution page
  8. Artur Jerzy Barczyk (California Institute of Technology (US))
    15/05/2013, 14:30
  9. Prof. Motonori Nakamura (NII)
    15/05/2013, 14:50
  10. Michael Ernst (Unknown)
    15/05/2013, 15:30
    - How much networking do we need? Can we deduce the future needs by looking at what we have used in the past? - Is it sufficient to be agnostic/reactive and use what's been offered to ATLAS, or would there be an advantage of being proactive? - Potential of Network-aware Applications in ATLAS - what would be the benefits and what are suitable areas?
    Go to contribution page
Building timetable...