Attending: Jens (chair+mins), David, Dan, Teng, Brian

Apologies: Sam, Matt


Brian, supplemented by David, gave a summary of the most important/interesting takeaway topics at the
WLCG workshop in Naples yesterweek.  See references to talks in the chatlog.

* Datalakes - big topic, more next week.
  - "federated" storage eg dCache which is to support CEPH (and Lustre); EOS (more next week); DPM with dynafed.
* Storage evolution again - expecting Great Leap Forward in data in 2026, superexponential, needs thought.
  - how will we do the storage?
  - how will we do the analysis?
  - technology watch lookahead (for both), and analysis and storage overlaps with some technologies like SCM, etc.
* Data preservation - good at bitlevel.  CMS preserved data required SL5 to make sense of it.
  - In particular, SLAC to cease supporting BaBar tape by 2021 - "only" 2PB - who will store it?
* How to transfer data (presentation by B Bockelman, so using scitokens)
  - also 3rd party copying with xroot, http
* CMS Dynamo similar to Rucio.
  - something called "Harvester" which we will need to revisit.
* Cost models, most sites don't have to account directly for every cost.  Towards understanding true costs of eg
  workflows, or tape vs disk (is access counted in the cost?)
* Summary talks Thursday; also live notebook summary of discussion.


Chatlog:

David Crooks: (04/04/2018 10:03:23)
https://indico.cern.ch/event/658060/contributions/2940554/attachments/1625111/2587608/WLCGDataManagementSummary.pdf
Brian Davies @RAL-LCG2: (10:05 AM)
https://indico.cern.ch/event/658060/contributions/2844782/attachments/1622746/2582912/ScienceGoalsWLCG-HSFworkshop2018.pdf
https://indico.cern.ch/event/658060/contributions/2889027/attachments/1622791/2583013/tech_market_BPS_Mar2018_v9pptx.pdf
https://indico.cern.ch/event/658060/contributions/2886720/attachments/1623465/2584401/Datalakes-WLCG-HSF-Naples2018-v2.1.pdf
https://indico.cern.ch/event/658060/contributions/2915947/attachments/1623301/2584548/DPHEP-WLCG-HSF-Collaboration-March2018.pdf
Daniel Peter Traynor: (10:20 AM)
all the data and software from the jade experiment from th e80s fits on to one cd
jens: (10:21 AM)
Just check if you can read it and whether the code still compiles...
Daniel Peter Traynor: (10:24 AM)
jade reanalysed ther data ~5 years ago when new NNLO jet theory came out. They managed to find the calabration data in former spokepersons office on printed sheets and typed them back into a computer
Brian Davies @RAL-LCG2: (10:25 AM)
https://indico.cern.ch/event/658060/contributions/2886727/attachments/1623451/2584571/WLCG_Transfer_Ecosystem.pdf
https://indico.cern.ch/event/658060/contributions/2917675/attachments/1623445/2584360/Dynamo_Naples_18.pdf
https://indico.cern.ch/event/658060/contributions/2905697/attachments/1624402/2586530/ResourceModel.pdf
https://indico.cern.ch/event/658060/contributions/2886775/attachments/1624460/2586300/Capability-Auth-Technical-Details.pdf
https://indico.cern.ch/event/658060/contributions/2940556/attachments/1625107/2587592/Frameworks_and_Infrastructure_-_sumary.pdf
https://indico.cern.ch/event/658060/contributions/2940579/attachments/1625117/2587621/Naples-Wenaus-WM-Summary.pdf
https://indico.cern.ch/event/658060/contributions/2940554/attachments/1625111/2587608/WLCGDataManagementSummary.pdf
David Crooks: (10:45 AM)
https://indico.cern.ch/event/658060/overview
https://indico.cern.ch/event/658060/attachments/1575251/2580134/go
Daniel Peter Traynor: (10:50 AM)
dcahe to support lustre as well