Enabling Grids for Escienc E Simulations and Offline

Enabling Grids for E-scienc. E Simulations and Offline Data Processing for the Auger Experiment Jiří Chudoba, Institute of Physics, ASCR and CESNET Jaroslava Schovancová CESNET www. eu-egee. org EGEE-II INFSO-RI-031688 EGEE and g. Lite are registered trademarks

Outline Enabling Grids for E-scienc. E • • Cosmic rays Pierre Auger Observatory Computing Model VO Auger Data Management Policy Auger Grid CPU Challenge Simulation Plans EGEE-II INFSO-RI-031688 VO AUGER 2

Cosmic rays Enabling Grids for E-scienc. E Paul Sommers, 2003 Sources of particles with E > 1020 e. V have to be within “GZK-sphere” (100 Mpc) Simon Swordy, 1996 EGEE-II INFSO-RI-031688 VO AUGER 3

Showers Enabling Grids for E-scienc. E • Primary particle interacts with atmosphere • Number of secondary particles is created • Secondaries interact again, and again, . . . • Typical shower 1020 e. V: 1010 particles at ground • Animation color code: blue: electrons/positrons cyan: photons orange: protons red: neutrons gray: mesons green: muons (10 -6 thinning) EGEE-II INFSO-RI-031688 VO AUGER 4

Pierre Auger Collaboration Enabling Grids for E-scienc. E • More than 250 Ph. D scientists from more than 60 institutions from 16 (+2) countries • > 400 authors Participating countries: Argentina, Australia, Bolivia*, Brazil, Czech Republic, France, Germany, Italy, Mexico, Netherlands, Poland, Portugal, Slovenia, Spain, United Kingdom, USA and Vietnam* * - associated countries EGEE-II INFSO-RI-031688 VO AUGER 5

Pierre Auger Observatory Enabling Grids for E-scienc. E • Southern observatory almost completed • Some results already presented Northern hemisphere (planned): Lamar, Colorado, USA EGEE-II INFSO-RI-031688 Southern hemisphere: Malargüe, Mendoza province, Argentina Lifetime of the observatory: 15 - 20 yrs VO AUGER 6

Detectors Enabling Grids for E-scienc. E Fluorescence telescopes: Number of telescopes: 24 Mirrors: 3. 6 m x 3. 6 m with field of view 30º x 30º, each telescope is equipped with 440 photomultipliers. Ground detectors: Covered surface: 3000 km 2 Number of detectors: 1600 Type of detector: Detector of Cherenkov radiation, each consisting of 12 000 litres of ultrapure water and equipped with 3 photomultipliers. Spacing between detectors: 1. 5 km. EGEE-II INFSO-RI-031688 VO AUGER 7

Event Reconstruction Enabling Grids for E-scienc. E EGEE-II INFSO-RI-031688 VO AUGER 8

Auger Computing Model Enabling Grids for E-scienc. E • Data recorded at the experiment site are copied to Lyon - CCIN 2 P 3 – slow line from Malargue, transfer via recorded DVD – EU funded project AUGERACCESS will improve connectivity • All Auger members have accounts at CCIN 2 P 3 • Data is analyzed remotely at CCIN 2 P 3 or copied and processed on PCs or small computing farms – users must install new versions of the offline sw – users check for availability of new data • CPU intensive simulations done at various sites and results copied to CCIN 2 P 3 – some sets are not at CCIN 2 P 3, available to only some users EGEE-II INFSO-RI-031688 VO AUGER 9

Expectations from Grid Enabling Grids for E-scienc. E • more computing resources available to more collaboration members • uniform way for simulation productions • easier access to data • tests with simulations were done using regional VOs (central Europe - VOCE, South. West) EGEE-II INFSO-RI-031688 VO AUGER 10

VO AUGER Enabling Grids for E-scienc. E • created in 2006 by the Czech group in cooperation with CESNET • CESNET provides and maintains central resources – LCG RB, g. Lite WMS, LB, UI, LFC, registration portal, VOMS server • VOMS server replica at FNAL • 5 sites from 4 countries support VO AUGER – we expect more sites after discussions during collaboration meeting at the end of April • EELA project considers AUGER as supported application – E-infrastructure shared between Europe and Latin America – www. eu-eela. org, IST-2006 -026409 – UNAM (National Autonomous University of Mexico) is setting up support for VO AUGER EGEE-II INFSO-RI-031688 VO AUGER 11

Data Management Enabling Grids for E-scienc. E • Data Management Policy document • 1 LFC server • All files must be registered there – orphan files can be deleted by administrators • LFC: – Users can use only directories /grid/auger/user/<username>/. . . • SE: – Files stored in a generated area are not allowed – -P option must be always used with lcg-cr and lcg-rep commands • Regular checks • Middleware tools for policy enforcement would be useful EGEE-II INFSO-RI-031688 VO AUGER 12

Small Productions Enabling Grids for E-scienc. E • Set of scripts for CORSIKA simulations – user defines number of jobs, lower and upper values of parameters – scripts do job submission, job monitoring, output retrieval and optional job resubmission – tested by users, not used for bigger productions so-far EGEE-II INFSO-RI-031688 VO AUGER 13

Auger Offline on Grid Enabling Grids for E-scienc. E • Auger offline code was tested on Grid – – CESNET in collaboration with partners from IJS, Slovenia installation done locally in directory accessible from WN binary relocatable tarball is being prepared separate My. SQL server with data needed by jobs set up for grid jobs • Atmosphere monitoring jobs – modified version of offline code – golden hybrid events reconstructed with varied atmospheric conditions EGEE-II INFSO-RI-031688 VO AUGER 14

Auger Grid CPU Challenge Enabling Grids for E-scienc. E • Purpose: – get realistic numbers for number of CPUs available – check stability of sites and infrastructure • • • Jobs about 1 CPU hour long Submit enough jobs to keep queues full All jobs submitted by 1 “common” user Phase 1: no warning to sites Phase 2: inform sites Timescale: 58 hours per phase (2 weekends) EGEE-II INFSO-RI-031688 VO AUGER 15

CPU Challenge - Phase 1 Enabling Grids for E-scienc. E EGEE-II INFSO-RI-031688 VO AUGER 16

CPU Challenge - Phase 1 Enabling Grids for E-scienc. E EGEE-II INFSO-RI-031688 VO AUGER 17

CPU Challenge - Phase 2 Enabling Grids for E-scienc. E EGEE-II INFSO-RI-031688 VO AUGER 18

CPU Challenge - Phase 2 Enabling Grids for E-scienc. E EGEE-II INFSO-RI-031688 VO AUGER 19

Auger Grid CPU Challenge - Results Enabling Grids for E-scienc. E • Phase 1 – 3 sites running jobs + 2 sites aborting all jobs – 160 (maximum), 90 (average) concurrent jobs running – 2152 (80 %) jobs done • Phase 2 – 4 sites – 220 (maximum), 100 (average) concurrent jobs running – 2250 (51 %) jobs done • Some problems were already solved • Observed issues – stability - WMS, sites EGEE-II INFSO-RI-031688 VO AUGER 20

Simulation plans for Grid Enabling Grids for E-scienc. E • Library of CORSIKA showers with EPOS model – – 8 energy bins, 7 angle bins, 100 events per bin 5600 proton showers 5600 iron showers 5 TB of simulated data • Timescale: 50% in November 2007 – similar production with faster model took 5 months on dedicated resources at CCIN 2 P 3 • Collaboration will consider to include more resources in the Grid if we demonstrate effective usage • Easy interoperability with OSG would be welcome – FNAL accessible through g. Lite CE EGEE-II INFSO-RI-031688 VO AUGER 21

The End Enabling Grids for E-scienc. E • Many thanks to members of Auger Collaboration for pictures and animations used in this presentations EGEE-II INFSO-RI-031688 VO AUGER 22