Attending: Jens (chair+mins), Sam, Steve, Brian, Dan, RobC, Pete Clarke, Matt, Teng, JohnH

0. Operational blog posts

   Glasgow - clean up of ATLAS data, some troubles with data not actually having been cleaned up and it needs
   to be deleted manually.

1. Following up from yesterweek's discussion, none other than Pete Clarke will join us this morning.  

   We are to start putting a paper together in preparation for GridPP6; unlike previous efforts, it should
   start specifically with the required software components - not hardware, not experiments' efforts,
   not operations.  Areas of interest are storage, data movement, relevant information systems, and including
   T1, not just T2s.  Development and research can be included in what is often a changing landscape and,
   while we don't write software as such, there is much testing of new stuff and often scripts and puppet
   scripts etc. are required, or interoperability testing.  An agile approach will be needed.  Steve points
   out the component based approach can help interoperation, testing, and reuse (particularly if they have
   standard, documented interfaces).  Users might want to bring their own cloud, as climate modellers do
   - we should worry about the technical side, not the costing of this.

   For each component, we should cover current status, future developments, what needs doing, and what will
   be the role of GridPP6.  As ever, we need to avoid duplication, work towards harmonisation and support
   hopefully unified WLCG experiemnts, yet not rule out models that support non-WLCG

   For example, the IPv4-IPv6 issues need to be covered - there are nine months for larger T2s to enable
   IPv6, but all sorts of problems arise as we've seen with dual stacked sites, or with CASTOR that doesn't
   support IPv6, and IPv6-only sites (or SEs) will also be interesting when they appear.

2. Last minute comments on the CHEP submission(s)?

   Sam will share his slides later today.  The submission looks at the evolution of T2s, so has a fair
   bit in common with the document we have been asked to draft.  Cache-only sites, distributed SEs,
   using common standard protocols in addition to, or instead of, the traditional HEP/grid ones, in the
   former case with a lightweight interface supporting the grid protocols (a la StoRM)

3. AOB

   The DOMA WG seems to have sprung back to life and is premeditating a meeting at CHEP; topics for which may
   include post-X.509 [client] authentication, xroot proxy cache, and anything else in the post-SRM world.

   Since most of the Usual Suspects are away next week, we will cancel next week's call.

   End of quarter - send all relevant stuff if you are Sam, Brian, Teng, or RobC (that we don't already
   know about)

   Reminder of the pre-GridPP workshop planned - DPM DOME hack day

jens: (04/07/2018 10:23:44)
http://storage.esc.rl.ac.uk/weekly/
Ste Jones: (10:23 AM)
One term used for this is "Component Based Software Engineering". I think the storage group does a lot of this, BTW. 
Brian Davies @RAL-LCG2: (10:26 AM)
https://ggus.eu/index.php?mode=ticket_info&ticket_id=135822
Ste Jones: (10:26 AM)
Uusually the term means classes/objects. But, at a higher level of abstraction, it can refer whole applications, clusters of applications, interoperability of systems, training, baseline formation, testing, best practise, communications. etc.
Since we don't do that much actual writing of code, what remains is, largely, high level component based software engineering, in my view.
Anyway, that's one way of looking at it...
In fact it is best called "Component Based _Systems_ Engineering, in our context.