Attending: Sam, Elena, Robert, Ewan, Mark, Govind, Dave, Steve, Jens Apols: Wahid, John H 1. Couple of things looking interesting at the GDB today; in particular, the DPM support issue seems to have made Progress(tm), as we have discussed by mail over the past week. 2. Elena provided some feedback from Monday's Jamboree. We will need to do a bit of follow up next week as well. Jeremy had kindly circulate the URL for the agenda page: http://indico.cern.ch/conferenceDisplay.py?confId=155075 Highlights covered today include: * The need for xrootd * testing arrangements for analysis jobs (WN access to storage) * Deploying HTTP - "ASAP" * "Intelligent networks" - Elena will send some stuff to the list. Easier to set up xrootd now than it used to be, in recent versions of DPM. Yaim probably doing a good job provided it has started from a clean install; if you already have manually configured stuff, it might be more of a problem - check your updated config. A problem reported earlier turned out to be wrong filenames; there may now be a handshake problem which is being investigated. 3. Anything else people should do before the Christmas break? Probably not a huge lot you don't know. Check your certs, that you have none expiring. Govind had a problem with PeCR, will check offline, talking to JK. Any (RFIO) client library problems in upgrade? May need to check before you turn off the light? 4. Big data - "anything bigger than you can handle" - for us, a few terabytes is small but for some users it's still big... Royal Society report on Big Data - DC contributed on behalf of GridPP, will circulate link. GridPP doing lots of big data stuff, but perhaps our data is more regimented than the general case, with our data models and structured files. Maybe we should look for case studies, and/or new users. Bioinformatics can be quite active and have lots of data - eg Alastair at Glasgow (Mark's chum). Gene sequencing work or discussions at Oxford and Imperial; and STFC have worked with WTSI (aka Sanger) storage folk. SKA - Shaun talking to them, also David Wallom. Jens coauthored report on NRENs future for ASPIRE on behalf of WLCG. SKA "bigger than HEP", but then HEP is bigger than HEP - Oxford grown from 6 TB to 800 TB in five years. When SKA get around to takign data, we may have the same size that they're expecting. CHEP next year will also have "big data" as a topic. Dave the dataset also illustrated the case well. 5. AOB CMS increasing reliance on xroot. Also ATLAS (FAX). Redirection (formely known as "federation") preprod rehearsals in January. Could we use indico (again)? Jens doesn't have a CERN account and doesn't like to have one but everybody else would like indico. Is EVO being phased out and if so will we have to use Vidyo? JANET are continuing UK support for (Se)evo(gh). Various people will check various mail and if successful mail the list. [10:02:14] Sam Skipsey ...aha [10:03:39] Robert Frank yes [10:04:43] Elena Korolkova And Chris Walker is still in CERN [10:04:56] Elena Korolkova indeed for GDB [10:04:56] Jens Jensen http://indico.cern.ch/conferenceDisplay.py?confId=155075 [10:05:40] Ewan Mac Mahon joined [10:05:50] Mark Norman joined [10:06:53] Govind Songara joined [10:10:51] Ewan Mac Mahon I'm not sure I'd call Oxford completely ready, it's sort-of working, probably. [10:11:00] Ewan Mac Mahon It's about to get stomped over by YAIM though. [10:11:48] Elena Korolkova FAX people are arranging testing and tutorials. [10:12:12] Elena Korolkova I'll send the talks to the storage lists. [10:12:40] Elena Korolkova I've missed the FAX talk. [10:17:16] Govind Songara David want to join via skype he need meeting id [10:18:07] Jens Jensen Sent the details to the list [10:21:33] Skype Bridge joined [10:22:28] Stephen Jones joined [10:26:43] Sam Skipsey Although we should say that Human Genomes are not that big, really, data-wise. [10:27:11] Ewan Mac Mahon No, but the plans seem to involve scans over lots of them to pick out commonalities. [10:27:42] Elena Korolkova Also atlas is moving from srm to Rucio and Rucio has a plugin for DMLite. [10:27:43] Sam Skipsey A human genome is about 3 GB. [10:27:55] Sam Skipsey So, they're nicely comparable to our AODs. [10:28:03] Ewan Mac Mahon And I think there might be some derived data. [10:28:13] Sam Skipsey (I guess they compress well, so probably they end up being a bit smaller) [10:28:25] Ewan Mac Mahon The Oxford folks were certainly thinking in terms of a few petabytes. [10:28:42] Sam Skipsey That's a lot of genomes, or bits of genomes, yes. [10:28:54] Ewan Mac Mahon And the last I heard the Welcome Sanger Institute had a ~16PB lustre, so they're got something on there. [10:30:55] raul lopes joined [10:33:33] Ewan Mac Mahon Someone should do a blog post on this tomato thing. [10:34:31] Ewan Mac Mahon Yes; it would be great to get better 'summary' stats on that. [10:36:52] Ewan Mac Mahon But the shape of the problem is very similar - the Sanger's banks of sequencers will be churning out data faster than they can deal with it on site, [10:37:12] Ewan Mac Mahon so they're about to run into the same sort of problem that we already have with the LHC [10:37:40] Ewan Mac Mahon They're going to need to get the stuff off site quickly and process it somewhere else. [10:38:24] Ewan Mac Mahon I think a lot of it's tricky, but if they can come up with something that doesn't have those problems, there might be scope for making some progress. [10:38:44] Ewan Mac Mahon But we can't know that without asking the bio folks. [10:43:49] Elena Korolkova I think if we want to keep meetins on EVO, then it's better not to use indico [10:44:01] Skype Bridge left [10:44:09] Stephen Jones left [10:44:45] raul lopes left [10:47:09] Robert Frank left [10:47:22] Elena Korolkova bye [10:47:25] Elena Korolkova left [10:47:34] Mark Norman left [10:49:42] Ewan Mac Mahon left [10:50:07] Sam Skipsey Jens: one other "big data" thing, which Mark reminded me of. [10:50:57] Sam Skipsey We're doing a pilot thing with the archeologists, doing high-res scans of small archeological finds, some 3d reconstruction, also some interconnection with the detector development people here at Glasgow PPE. [10:52:08] Govind Songara left [10:52:21] Jens Jensen thanks sam