Attending: Jens (chair+mins), Sam, RobC, Teng, Winnie, Gareth, Matt

Apologies (and attempts to join): Sam

0. OBP

   No operational issues.  New blog post from Rob!

1. GridPP last week - summary, highlights, conclusions

   Hackathon - considered useful.  Lancaster bravely volunteered to be the first to switch to DOME (thanks Matt.)
   Clear that infrastructure needs to be all IPv4 or all IPv6 or all dual stacked.

   Security Wedn. morning, not directly storage related.

   Most of discussion was about data lakes (the CERN variety).  What are the minimal requirements for running a data lake?
   Is RUCIO data lake aware and if so does it work a bit like the Data Carousel?

   T2 evolution - how to progress, in large steps or small steps...
   Like, would there be but a single endpoint for - the T2, or for all the GridPP T2Ds.  Additional layers must not be experiment specific.
   E.g. if we had five large T2Ds, how might we federate them?

   - Through a shared storage solution like dCache (a la NDGF), or the EOS variation.
   - Through a shared filesystem like DEISA used to have (but notably PRACE chose not to have, so probably not a great idea)
   - Federation at protocol level, like dynafed or federated xroot (cf DOMA).  Plus the caches of course.

2. If there's time I will say a few words about a talk I am giving Friday about archiving data

   Workflows for archiving - Gareth noted Markus Schulz's working group looking at reproducibility which aligns with our 'output' class of data.

3. AOB