DOMA / ACCESS Meeting

Europe/Zurich
513/1-024 (CERN)

513/1-024

CERN

50
Show room on map
Frank Wuerthwein (Univ. of California San Diego (US)), Ilija Vukotic (University of Chicago (US)), Markus Schulz (CERN), Stephane Jezequel (LAPP-Annecy CNRS/USMB (FR)), Xavier Espinal (CERN)

People on Vidyo: Alessandra Forti, Alexei Klimentov, Andrew Melo, Bo Jayatilaka, Carlos Perez Dengra, Diego Ciangottini, Doug Benjamin, Duncan Rand, Frederic Derue, Frank Wuerthwein, Gonzalo Merino, Ilija Vukotic, Justas Balcas, Laurent Duflot, Marcelo Vilaca, Markus Schulz, Nikola Hardi, Nikolai Hartmann, Oxana Smirnova, Riccardo Di Maria, Stephane Jezequel, Xavier Espinal

Questions to site admin:

Evolution of internal organisation induced by larger capacity in 2028
   - Foresee to run with same manpower at HL-LHC scale ? Support non-LHC VO ? Same size ? 
   - Large local CPU capacity ? 
Feedback on datalake : remote access, protection with xcache in front of WN or in network 
   - Feeling about remote access can become much larger than controled/predictable local one ?
   - Availability of larger bandwidth dedicated to your site ?
   - Any additional feedback on datalake proposal ? 

 

* Frank Wuerthwein (Univ. of California San Diego (US)) - Impact of Data Lake Model on total cost of ownership: US CMS T2

  • please see slides for details and numbers
  • data loss due to power outage mainly
  • reference points: 10k hyper threads, 5PB of usable total disk space
  • data volume of RAW vs SIM is 1:1
  • CMS tapes only at FNAL in this scenario
  • willing to have Rucio controlling the buffer space for processing workflows
  • 450 Gbit/s archival from FNAL
  • erasure encoded CEPH: with at least 3/4 disk security based on the CEPH storage (namespace), not on machines
  • CEPH slides presented in OSG AHM: https://indico.fnal.gov/event/22127/contributions/194938/
  • Xavi: Excellent discussion today. Unfortunately, I need to dash now. I encourage to present an extract of this presentation at the WLCG HSF storage workshop. These are the right Qs to expose. Bye.
  • action to organise meetings with people performing data popularity studies and/or key-contacts

 

There are minutes attached to this event. Show them.
    • 17:30 17:35
      Introduction 5m
    • 17:35 18:15
      Impact of Data Lake Model on total cost of ownership : US CMS T2 40m
      Speakers: Frank Wuerthwein (UCSD), Frank Wuerthwein (Univ. of California San Diego (US))