The CNAF Tier 1 center XXIV HTASC Meeting

The CNAF Tier 1 center XXIV HTASC Meeting Pisa, 12 June 2003 Guido Negri – INFN-CNAF, Bologna Guido Negri – Pisa, 12 June 2003

General Informations • CNAF is the INFN National Centre for Research and Development in Telematics and Informatics Technologies. • Originally was dedicated to the measurements of the Bubble Chambers photos. • Hosted by the Physics Dept. of the University of Bologna. • Personnel: 17 staff + 17 time limited contracts. • Total annual budget around 1. 7 M€ + Projects funds Guido Negri – Pisa, 12 June 2003

CNAF Activities • General INFN services § § § AFS, DNS, WWW, News and Mail… Video Conference: H 320 & H 323 MCU Hosting of training courses (C, C++, ODB…) • Projects § § TIER 1 Regional Center GRID: Data. Grid, Data. TAG, INFN-GRID, EGEE, LCG, Grid. it High Bandwidth Networks: Optical Networks, WDM, GARR-G Pilot Testing of new technologies Guido Negri – Pisa, 12 June 2003

TIER 1 – General Considerations TIER 1 is meant as a computing facility for INFN HNEP community CNAF will be a multi-experiment TIER 1 (ATLAS, CMS, LHCb, ALICE, VIRGO and later CDF) Aims • providing experiments with computing resources • support to TIER 2 s and TIER 3 s • coordination with TIER 0, other TIER 1 s and TIER 2 s INFN-TIER 1 is only a prototype • TIER 1 becomes fully operational: end of 2003 • end of project Phase I: beginning of 2004 Guido Negri – Pisa, 12 June 2003

TIER 1 Services • • Computing servers (CPU farms) Access to on-line data (disks) Mass storage / tapes Broad-band network access and Qo. S System administration Database administration Experiment specific library software Helpdesk Guido Negri – Pisa, 12 June 2003

Issues Technical staff Recruiting & Training Resource management Minimization of manual operations Sharing of resources (network, CPU, storage, HR) among experiments Resource use optimization Compatibility between tests and production activity Technological tests for Tier-1 Prototype phase (LHC experiments) Production phase (VIRGO, Ba. Bar, CDF II, etc. ) Integration with grid framework (EDG, EDT and LCG) Interoperation Common tool development and test Guido Negri – Pisa, 12 June 2003

Computing Resources Computing servers (CPU farms) • 150 (soon ~320) 1 U bi-processors Pentium III/IV 8002400 MHZ (IBM, DELL, Super. Micro) • System installation and administration – Linux Red. Hat (6. 2, 7. 3) – Experiment specific libraries and software – LCFG, Remote access to KVM. Access to on-line data (DAS, NAS, SAN) • • ~35 TB (soon to become >70 TB) Study of large file system solutions (GFS, GPFS) SAN on WAN tests (collaboration with CASPUR) Test of several HW technologies (EIDE, SCSI, FC) Guido Negri – Pisa, 12 June 2003

Network (1) New GARR-B Backbone with 2. 5 Gbps F/O lines already in place (soon to pass to 10 Gbps). CNAF-TIER 1 access is now 1 Gbps. Gigapop is collocated within INFN-TIER 1. Many TIER 2 s are now 155 Mbps. International Connectivity via Geant: 10 Gbps access in Milano and 3 x 2. 5 Gbps links of Geant with US (Abilene) already in place Guido Negri – Pisa, 12 June 2003

Network (2) GARR-B GEANT 10 2. 5 Gb p 2. 5 ps 2. 5 Gb ps BO CT 155 TIER 1 1 Gbps ps s bps 2. 5 Gb ps Gb MI 2. 5 Gbps s 2. 5 PI s Mbp p Gb TO 155 PD s p Gb RM b G 5. 2 M Guido Negri – Pisa, 12 June 2003 CNAF

TIER 1 Network Bo 12 KGP GARR-B 1 Gb/s SSR 2000 R&D Catalyst 6500 R&D Sez. Di Bologna Matrix M 5 SSR 8600 Gigabit Switch Router Disk-serv-CMS Farm. SWG 1 Extreme 7 i NAS 4 NAS 2 NAS 3 Guido Negri – Pisa, 12 June 2003 8 T F. C.

Storage Mass storage/tapes • Storage. Tek library with 9840 (30 tapes) and LTO drives (150 tapes, 100 GB each) • CASTOR as front-end software for archiving – Direct access for end-users – Oracle as back-end § New library with 2000 -5000 tapes capability in September Disk storage • 8 TB Fibre Channel (DELL) • 2 TB Fibre Channel-EIDE (AXUS) • 2 TB SCSI (Raitec) used as staging for CASTOR • 16 TB NAS based on FC (Procom) • 2 TB NAS IDE Guido Negri – Pisa, 12 June 2003

CASTOR at CNAF STK L 180 2 drive 9840 LEGATO NSR (Backup) SCSI LAN Robot access via SCSI ACSLS SCSI 4 drive LTO Ultrium CASTOR 2 TB Staging Disk Guido Negri – Pisa, 12 June 2003

Developed at CNAF – Monitoring by Felice Rosso Farm monitoring tool • series of light scripts (bash, Perl) monitoring the status of a computing farm (CPU Load, disk usage. . . ) • • • fast: <1 sec for 100 machines tainted: no shell allowed no multi-thread (no hacking allowed) remote monitoring soon to come more efficient tool (Python) Guido Negri – Pisa, 12 June 2003

Developed at CNAF – Resources Database by Barbara Martelli Aims • Centralized repository for hardware informations • Monitoring • Alarm • Configuration facilities Based on • Postgre. SQL • PHP interface • Web interface Managed Data • Batch informations • hardware informations • Technical assistance • Software configuration • Network configuration Guido Negri – Pisa, 12 June 2003

Conclusions INFN TIER 1 is offering an experimental service… • VIRGO, CMS, ATLAS, LHCb, ALICE • Data. Grid and Data. TAG test-beds …but we are still in a test phase • Study and tests of technological solutions Main goals of the prototype are • train people • adopt standard solutions • optimize resource usage • integration with the grid Operational phase forseen starting from end of 2003 Guido Negri – Pisa, 12 June 2003