Towards deploying a production interoperable Grid Infrastructure in the U.S. Vicky White U.S. Representative to GDB

May 18, 2004GDB - US2 Grid Deployment and Data Challenges u US CMS participated in DC04. Generally a lot of “glue” development and operational work u simulation production benefited from 50% increase in resources through opportunistic use of Grid3. u US ATLAS DC2 is deployed and ramping up production. u Don Quixote isolates application from different Replica Location Service infrastructures. u End to end “virtual data”-based US production on Grid3 u US Edge Computing deployed at CERN to facilitate Tier-0 to Tier 1 data movement

May 18, 2004GDB - US3 Daily data transferred to FNAL (Y. Wu, CPT week)

May 18, 2004GDB - US4 Fermilab Dcache nodes network traffic (Y. Wu, CPT week)

May 18, 2004GDB - US5

May 18, 2004GDB - US6 Progress on Storage Services u dCache/SRM used on US CMS Tier-1 and Tier-2s for DC04 analysis. u dCache/SRM being evaluated at BNL u LBNL DRM being tested in VDT. LBNL HRM (SRM) is use for STAR production data movement between BNL and LBNL. Plans for interoperability tests between LBNL HRM and dCache at BNL. u DESY/Fermilab collaboration providing support for LCG deployment of dCache/SRM u Separation of SRM layer underway at Fermilab. Collaboration with University of Wisconsin to enable NeST back end. u Starting to see other uses: dCache in use for Biology applications on at Wisconsin; Interest from BTeV at Vanderbilt. u In the process of sorting out the detailed issues of license of source code and ongoing development and support.

May 18, 2004GDB - US7 Virtual Organization Services u EDG-VOMS in use on Grid3. u VOX VO registration service (VOMRS) in test at BNL for Nuclear Physics Experiments. u VOMRS being evaluated by LCG for short term interim use. u We are ready to collaborate further on request. u Authorization Privilege Management joint project across ATLAS, CMS, Run II, Facility storage services. u Plan for deployment on US grid by the end of the year. u Computer scientist technology lead is Markus Lorch from Virginia Tech u We are ready to collaborate further on request.

May 18, 2004GDB - US8 Computation Services u US CMS is testing the sharing and publication of compute resources to Grid3 and LCG. u Job execution Policy provided by the batch queue interface and supports sharing of the CE between local batch user use, Grid3 and LCG jobs.

May 18, 2004GDB - US9 Security Services u US people (Dane Skow and Rich Baker) are Participating in the LCG security group. u Comments provided on User Registration Document. u The result seems reasonable to us. u In the US possible response to Teragrid site security incident seems to be a move towards the KCA-Like model deployed at Fermilab u LCG/EGEE security group  need a broader group to evolve policy and technology choices u Aware of security aspects of ARDA prototype, Dane Skow talking with Miron.

May 18, 2004GDB - US10 Software Services u Pacman3 in use for US ATLAS DC2. u Pacman3 evaluation planned soon for VDT. u Boston U & VDT agreement on support and maintenance of Pacman near.

May 18, 2004GDB - US11 Grid3 Status and Evolution u Grid infrastructure remained stable to completion of CMS DC04 and initial use for reconstruction. u Little operational support load, see VO jobs rise and fall at level of several 100. u Split Grid into development and operational grids. u Run II converting to use of VDT over next couple of months (lesson of running experiments - would benefit from an SGI port)

May 18, 2004GDB - US12 Stability of Grid3 infrastructure u Currently SDSS doing a 2 week ~40,000 job, individual investigator science run “opportunistic science” The coadd itself consists of transferring all the images of a location in the sky out to a grid node, remapping the images into a standard coordinate system, and peforming a variance weighted average. There are roughly 7000 fields, in 5 colors, so there are 35,000 jobs to be run per coadd (we are doing two to explore parameter space). Each job takes about minutes to run. After the coadd itself, we run the object detection and measurement code, Photo, on the output, and bring everything back to Fermilab.

May 18, 2004GDB - US13 Development and Operational parallel environments u Enables testing of VDT and use of earlier versions of VDT in parallel

May 18, 2004GDB - US14 Participation in ARDA, EGEE/Middleware? u US LHC funds Miron Livny on ARDA/EGEE middleware u US CMS/CMS working on end to end ARDA prototypes for the fall. u US CMS will participate in CCS milestone for analysis across grids. u NSF funds VDT/EGEE joint effort. u Wisconsin certification testbeds for national NMI testing, VDT, LCG and EGEE middleware... etc. encourage coherence across the middleware versions. u PPDG helps fund ATLAS ARDA representative. u PPDG extension proposal is funded. Enables some immediate attention to Open Science Grid needs.

May 18, 2004GDB - US15 Open Science Grid u Open Science Grid is the Roadmap in the US for ensuring all of our Grid efforts, including and in particular the LHC ones, come together towards a coherent and sustained Grid infrastructure that u Provides computing for LHC science – US contribution to LCG u Is open to other experiments and other sciences u Will work with, collaborate with and interoperate with the Grid infrastructure provided through EGEE u Open Science Grid consortium is being formed - hope to come out of this weeks joint Trillium Steering, Laboratory Facility and Experiment software and computing management meeting with more formal structures in place. (Not a project) u Open Science Grid “Technical Groups” scope explicitly includes cooperation with the EGEE/LCG peers groups.