People on Vidyo: Alexei, Andrea Sciaba, Andreas Petzold, Bo Jayatilaka, David Michael South, Derek Weitzel, Doug Benjamin, Elizabeth Sexton-Kennedy, Frank Wuerthwein, Gonzalo Merino, James William Walder, Johannes Elmsheuser, Laurent Duflot, Nikolai, Oxana Smirnova, Riccardo Di Maria, Stephane Jezequel, Xavier Espinal, Xin Zhao
Stephane Jezequel (LAPP-Annecy CNRS/USMB (FR)) - ATLAS Repro Challenge HL-LHC:
- Inputs from:
- proposal from F. Wuerthwein to build reprocessing challenge based on HL-LHC numbers (10 PB/day)
- references for ATLAS numbers
- Run-2: numbers from last Run-2 reprocessing campaign (presentation at next GDB)
- HL-LHC : preliminary numbers from CDR
- Caveats
- ATLAS reprocessing: highly distributed by composition of ATLAS Grid
- Some infrastructure requirements not listed: size of buffer
- Frank computed numbers on a less precise way than Stephane (quoting Frank)
- The scenario under consideration is that reprocessing could overlap with data taking
- Clarification: for transfers, numbers refer to setup of 1 Rucio (dedicated instance) + 3 FTS servers (not sure if dedicated instances)
Discussion on Future Data Challenge
- Numbers (staging GB/s) from all the experiments are needed to estimate the impact on T1s
- Libraries are typically added/replaced every 2/3 years - adding capacity, etc.
- Several T1s are already asking requirements from experiments - funding can be organised up to 5 years
- Proposal for CMS and ATLAS to have these numbers from sites as a planning exercise
- Action on writing a document to have some numbers to work on - iterative process with T1s
- Aiming to describe "a" scenario by 2020 with definition of exercise
- This document is not aimed to be used officially for funding agencies
- To be checked with WLCG wrt LHCC if there is something - maybe not at the level wanted
Next meeting in 2 weeks, where it will be discussed the following.
Xavier's document:
- The document uploaded has been shown several times in DOMA General
- Aimed to start a discussion for the planning of data challenges with some metrics and timelines, according to the brief feedback obtained from the HL-LHC reviewers
There are minutes attached to this event.
Show them.
