Present (in approximate joinological order)
	John B
	Jens
	Robert
	Brian
	Govind
	Elena
	Gareth
	Steve
	Sam
	David
	Ewan
	Duncan
	Chris
	Matt
	Pete

Apols:
	Wahid


1. FTS

FTS3: Being used by CMS for their UK transfers.  SRM transfers with FTS3
considered quite solidly debugged.

Atlas have done some transfers, and are seeing generally high transfer
rates but overall low (or shall we say less high) throughput with lots of
small files.  Planning on doing transfers with the following lucky T2s:
Manchester, ECDF, RALPP, and QMUL.

What's the endpoint compatibility?  Could probably do SRM from/to GridFTP
easily, as it should still be able to use 3rd party copying.  However,
how would we transfer to/from something else, like HTTP(S) or xroot?
Would it need to be piped through the FTS server or something similar?

What can we learn from the logs?  Should check...

Could we use FTS instead of GO?  Sure, but may require some user
education.  Would be advantageous for some types of scenarios.  We would
not get the desktop-based transfer with the lightweight desktop client,
so GO wins that one...

2. iRODS

iRODS is a integrated rule oriented data service or system or something
with S.  Also supports GridFTP (via a module called Griffin) but this
may not give the full functionality that you'd get with the command
line interface.

The main access method is via username/password; users will run iInit
and can then run "i-commands" similar to Unix ones, ie iLs, iMv, iRm.
There are also iPut and iGet for upload/download (internally using a
proprietary protocol), and additional commands for metadata.

T2K have an iRODS instance at QMUL.  EUDAT is also based on iRODS.


[10:02:03] John Bland I'm better than my mic is
[10:02:14] Jens Jensen Ah, maybe that's why I didn't hear  
[10:04:09] Jens Jensen http://indico.cern.ch/getFile.py/access?contribId=16&sessionId=2&resId=0&materialId=slides&confId=246453
[10:21:47] Elena Korolkova Upto date transfers plots is there http://dashb-atlas-data.cern.ch/dashboard/ddm2/#tab=transfer_plots
[10:31:28] Ewan Mac Mahon It would be amusing simply to be able to stash PP data on EUdat.
[10:32:23] Ewan Mac Mahon The other possibility is that we might be able to get the non PP people to pick up FTS3 for their transfers even if they're never going near PP hardware.
[10:42:04] Jens Jensen http://www.stfc.ac.uk/1263.aspx
[10:42:47] Jens Jensen http://zuserver2.star.ucl.ac.uk/dirac/Resources.html
[10:43:55] Ewan Mac Mahon They don't though.
[10:44:16] John Bland the cambridge cluster, at least, has a reasonably large lustre system at the back end but I've not seen anything about transfers, I can check
[10:45:08] John Bland we just use scp, but our data is miniscule
[10:45:39] Sam Skipsey The discussion in the meeting was that they'd tried scp, and were interested in gridftp when I mentioned it.
[10:45:48] Sam Skipsey but there was basically nothing about that since that meeting
[10:45:56] Jens Jensen scp isn't great for high performance data moving
[10:47:22] Ewan Mac Mahon IMPACT!
[10:47:32] Ewan Mac Mahon It's apparently very important.
[10:48:07] Jens Jensen Yes - impact is important. Very.
[10:55:12] Pete Gronbech The AAA project does keep getting put to STFC so they know it's there if a pot of money comes along.
[11:01:52] Jens Jensen Ewan I found your cert - tse25 - will be signed today.
[11:02:09] Ewan Mac Mahon There should one one called 'lhcb-vac' as well;
[11:02:17] Ewan Mac Mahon the RA apparently approved it on Monday.
[11:02:25] Ewan Mac Mahon Which is the more interesting one.
[11:02:26] Jens Jensen Curious