Main parameters of Russian Tier 2 for ATLAS

Main parameters of Russian Tier 2 for ATLAS Russia-CERN JWGC meeting 09. 10. 05 A. Minaenko IHEP (Protvino)

Main Tier-2 tasks • Tier-0, Tiers-1: storage of all data types (Raw, ESD, AOD, SIM), event reconstruction, ESD and AOD production, main bulk of calibration • The main Tier-2 task is providing facilities for physics analysis using mainly AOD, DPD and user derived data formats • It includes also development of reconstruction algorithms using subsets of ESD and Raw • All the data used for analysis should be stored on disks and some unique data (user, group DPD) to be stored on tapes also as well as previous AOD/DPD versions

Main Tier-2 tasks • The second important task is a production and storage of MC simulated data • The ATLAS computing model implies that simulated data are stored mainly in Tiers-1 but Russian Tier-2 proposes to take the responsibility for the locally produced MC data • This implies that the proper modes of access to the data to be granted and the data should be backed-up on the tapes • Participation in calibration/alignment activity can be considered if there will be interest from Russian groups

Data taking conditions and data sizes • Luminosity – – – • • • 0. 5*1033 cm-2 sec-1 in 2007 2. 0*1033 cm-2 sec-1 in 2008, 2009 1034 cm-2 sec-1 in 1010 and after Event rate 200 events/sec Event numbers 109 in 2007, 2*109 each next year Raw 1. 6 MB/event ESD 0. 5 MB/event AOD 0. 1 MB/event Raw. Sim 2. 0 MB/event Reconstruction 15 k. SI 2 k*sec/event Simulation 200 k. SI 2 k*sec/event Analysis 0. 5 k. SI 2 k*sec/event Luminosity increase in 2010 leads to the event size increase by 50% and reconstruction time by 75%

Evolution of Raw, ESD, AOD sizes in ATLAS

Cost table 2006 2007 2008 2009 2010 2011 2012 Disc (CHF/GB) 3. 14 1. 96 1. 22 0. 77 0. 48 0. 3 0. 19 ATape (CHF/GB) 2. 5 1. 25 0. 9 0. 6 0. 4 CPU (CHF/SI 2 k) 0. 67 0. 44 0. 3 0. 19 0. 14 0. 09 0. 06 Used lifetimes for the different types of resources: • CPU - 3 years • Disk - 4 years • Tape - 5 years Commissioning: • 80% of the resources 2007 are moved to 2006 • 30% of the resources 2008 are moved to 2007

Resources for simulation • Number of events – 20% of real data • Contribution of Ru. Tier-2 to total statistics of all MC data – 10% • Efficiency of CPU usage – 85% • All types of data to be saved on automated tapes • All ESD, AOD and 20% of Raw simulated data to be kept on disks permanently

Tables with simulation resources Simulation resources evolution 2006 2007 2008 2009 RAW (TB) 0 40 120 ESD (TB) 0 10 30 AOD (TB) 0 2 6 Disc (TB) 16 32 60 Tape (TB) 42 83 156 CPU (k. SI 2 k) 128 209 321 2010 200 50 10 100 260 321 2011 320 80 16 160 416 321 2012 440 110 22 220 572 321 560 140 28 280 728 321 Simulation resources yearly increase 2006 2007 2008 2009 Disc (TB) 16 16 28 Tape (TB) 42 42 73 CPU (k. SI 2 k) 128 80 112 Disc (k. CHF) 50 31 34 Tape (k. CHF) 104 52 66 CPU (k. CHF) 86 35 34 Total (k. CHF) 240 119 133 2010 40 104 128 31 62 24 118 2011 76 156 80 36 62 11 110 2012 76 198 112 23 79 10 112 88 198 128 17 79 8 104

Evolution of disk, tape, CPU (k. Si 2 k) simulation resources at Ru. Tier-2

Yearly costs of added disk, tape and CPU simulation resources at Ru. Tier-2

Resources for physics analysis • Number of active users – 50 • 5% of total Raw data and 10% of ESD are permanently kept of disks to be used for algorithm developments and analysis • Volume of group DPD data is equal to 50% of AOD data • Volume of user data – 1 TB per user for 2008 data only and varies proportionally to events number • CPU power needed to analyze 2008 data only by one user – 15 k. SI 2 k • Total CPU power is proportional to the number of accumulated events and to the number of users: float CPU = Nuser*(Nev_year[i]/Nev_year[2008])*CPU_user_2008;

Evolution of disk, tape, CPU (k. Si 2 k) analysis resources at Ru. Tier-2

Yearly costs of added disk, tape and CPU analysis resources at Ru. Tier-2

Evolution of disk, tape, CPU (k. Si 2 k) total resources at Ru. Tier-2

Yearly costs of added disk, tape and CPU total resources at Ru. Tier-2

Summary • Ru. Tier-2 for ATLAS – distributed computing centre shared with other LHC experiments • Expected parameters for 2008 are – Disks 830 TB – Tapes 230 TB – CPU 1450 k. SI 2 k • Provides facilities for physics analysis of data by 50 active users and MC simulation of 10% of ATLAS simulated data • Important feature – linear rise of practically all resources proportionally to the accumulated statistics • Commissioning: 80% of 2007 resources are moved to 2006, 30% of 2008 resources are moved to 2007 • Institutes: BINP (Novosibirsk), IHEP (Protvino), ITEP (Moscow), JINR (Dubna), PNPI (Gatchina), SINP (Moscow)