1. HEPiX Fall 2009 Highlights Michel Jouvin LAL, Orsay jouvin@lal.in2p3.fr http://grif.fr November 10, 2009 GDB, CERN

2. 2 HEPiX Fall 2009 Highlights 10/11/2009 HEPiX at a Glance ~20 years old informal forum of people involved in “fabric management” in HEP -No mandate from any other body -More sysadmins than IT managers -Open to anybody interested -http://www.hepix.org : archive of all meetingshttp://www.hepix.org Main activity is a 1-week meeting twice a year -60-100 attendees per meeting, “stable” core -Focus : exchange of experience and technology review Mix of large and small sites, many involved in WLCG -Better understanding of each others -Each one benefit of the others -Most of the sites involved in grid computing “On-demand” working groups -Currently distributed file systems -Benchmark WG closed but activity continuing

3. 3 HEPiX Fall 2009 Highlights 10/11/2009 HEPiX Healthy Last meeting 2 weeks ago in Berkeley (LBNL) -70 people, many Europeans, all DOE labs http://indico.cern.ch/conferenceOtherViews.py?view=standard&conf Id=61917http://indico.cern.ch/conferenceOtherViews.py?view=standard&conf Id=61917 -Very active contribution from LBNL/NERSC Link with other communities and other types of computing Attracting new sites in HEP… -PIC, Lisbon (LIP T2), CNAF… most of them now regular attendees -Next Spring meeting in Lisbon! … and out of HEP, also coming regularly -Astrophysics, Cornell in Berkeley -Genomics centers, Univ. of Gand in the past meeting -Generally somebody who spent some time in a HEP lab Want to return to Asia in Spring 2011

4. 4 HEPiX Fall 2009 Highlights 10/11/2009 Main Topics Main topics are always the same but main focus changes at each meeting -HEPiX value is the periodic report on work in progress -Umea: focus on virtualization Main tracks -Site reports : very valuable part, update on changes, give a « picture » of what’s happening aroud the world -Scientific Linux status and future directions -Data Centers : cooling, power consumption, … Less active, projects in building phase -Storage : convened by File System WG -Virtualization -Benchmarking : still active despite the end of the WG -Security and networking -Monitoring tools -ITIL: interesting reports from KIT, FNAL, CERN Diverse implementations, a process rather than receipes

5. 5 HEPiX Fall 2009 Highlights 10/11/2009 Virtualization A track at each meeting for more than 2 years… -Initial focus was mainly on service consolidation Coverage extended to virtualized environments for applications -Virtualized WN: integration with batch schedulers -Image management and resource allocation (openNebula) -CERNVM: very minimal and generic VM approach -Grid and clouds (eg. StratusLab) Progress since Umea meeting -Several sites reported advanced or production projects No performance penalty except for disk I/O: need for direct access rather than staging -US DOE launched a R&D project about clouds: Magellan LBNL+ANL: $US32M to implement a “scientific cloud”, evaluate potential benefit and scientific vs. commercial cloud Investigate efficient and reliable access to data Will include Infiniband clusters

6. Virtualization WG Tony Cass gave a talk: « a vision for virtualization in WLCG » -See next talk… Based on discussions in the last years and recent progresses, proposed a 5-step roadmap -Step 1: WN in a VM, image selected by the user -Step 2: sharing of images between sites, trusted method for image distribution -Step 3: inclusion of experiment SW in images Decided to create an HEPiX WG (chaired by Tony!) to share experience and make « rapid » progress -Build on success: CNAF, CERN… -Link with projects on grid/cloud complementarity -Report at each meeting -Not a forum to discuss with experiments: GDB for this Complementary: need to get sites in to make progress… 6 HEPiX Fall 2009 Highlights 10/11/2009

7. 7 HEPiX Fall 2009 Highlights 10/11/2009 File Systems WG Set up 2 years ago with a mandate to review distributed file systems technologies Work on a voluntary basis with 2 objectives -Benchmarking activities with realistic and diverse use cases: LUSTRE outperforming (2x) all other solutions in every case so far. -Share experience and expertise with new technologies Producing a report at almost each HEPiX meeting -No formal report this time… but probably 1 in Lisbon -New hot topic: Potential AFS + LUSTRE combination First tests of NFS 4.1 by dCache (presentation at LBNL) Discussion between European large labs and AFS gatekeepers for an AFS development program funded by HEP

8. HEP-SPEC06 Name of the benchmark defined by HEPiX benchmarking WG -Implementation in WLCG discussed by a follow-up GDB WG led by G. Merino Request from « EGEE » to rebrand the benchmark name to make it more acceptable by other communities -HEPiX considered the owner of the name and asked for agreement -Proposal: GRID-SPEC06, HS06… « HEPiX Board » had a long discussion and could not agree on the change -HEPiX is convinced, based on work by the WG, that there is no universal benchmark -If rebranded, others will be legitimate to ask for adjustments -Solution: allow a site to publish multiple units matching the benchmarks of VOs it supports 8 HEPiX Fall 2009 Highlights 10/11/2009

9. 9 HEPiX Fall 2009 Highlights 10/11/2009 Miscellaneous iSCSI evaluation at CERN -Follow-up on presentation at Umea -Very attractive performance for small I/O -“Inexpensive” redundancy using mirror of iSCSI devices served by different servers Allow to use “cheap” Storage-in-a-Box solutions for disk servers -Alternative to FC ? Status report of RAL migration to Quattor for managing the T1 -Steep learning curve as usual… but successful! -First T1 to use Quattor + QWG (YAIM replacement) -“Not necessarily easier to manage things but easier to keep things consistent” (I. Collier, project manager for Quattor migration)

10. 10 HEPiX Fall 2009 Highlights 10/11/2009 Conclusions HEPiX is a very « useful » forum opened to any site interested -Complementary to GDB, focused on fabric management rather than grid services -No formal membership : just register to the meeting.. Next meeting in Lisbon, April 19-23 -Look for announcement soon at http://www.hepix.orghttp://www.hepix.org -Ask to be registered on HEPiX mailing list (low volume…) -Next fall meeting in Cornell Material produced is mainly presentations during the workshops -Look at agendas if interested by a presentation. Start at http://www.hepix.org http://www.hepix.org

11. Quattor Workshop Last workshop held in Bruxelles last week -25 people, 15 sites, developers and users (but developers are users too!) Quattor adoption (slowly) growing -With RAL, now T0 + 3 T1 (CNAF and NIKHEF) RAL is the only « pure-QWG » T1 Other T1s planning to reuse more of the community effort known as « QWG templates » MS more and more active and now the biggest site -Managing 20K nodes with Quattor -Contributing a lot of components with a focus on scalability Now represent a significant part of the community effort Preparing QUEST proposal to be submitted to INFRA-2010-1.2.1.3 (MW and repositories) -Integrating with virtualisation and clouds is 1 major goal 11 HEPiX Fall 2009 Highlights 10/11/2009