Southwest Tier 2 Southwest Tier 2 Computing facility used by the - - PowerPoint PPT Presentation
Southwest Tier 2 Southwest Tier 2 Computing facility used by the ATLAS experiment Funded by the National Science Foundation as a subcontract to the U.S. ATLAS Operations program. A consortium of the University of Texas at Arlington
➢ Computing facility used by the ATLAS experiment ➢ Funded by the National Science Foundation as a subcontract to
➢ A consortium of the University of Texas at Arlington (UTA) and
➢ One of four Tier 2 facilities in United States of America ➢ There are forty Tier 2 facilities worldwide
➢ The LHC is a ~17 mile
➢ Accelerates beams of
➢ Collides the beams at
➢ Current Run II provides
➢ The size of a seven
➢ 90+ million data
➢ 40 million events
➢ 500-600 events per
➢ 4 megabytes per
➢ 2.5 GB/s stored to
➢ Tiered Model with Tiers 0, 1 ,
➢ Tier 0 is central facility
➢
CERN
➢ Tier 1 is a national facility
➢
Brookhaven National Laboratory (BNL) for the USA
➢ Tier 2 is a regional facility
➢
USA Tier 2's
➢
Northeast Tier 2
➢
ATLAS Great Lakes Tier2
➢
Midwest Tier 2
➢
Southwest Tier 2
➢ Tier 3 is an institutional
➢
University or group cluster
➢ Provides CPU cycles for:
➢
User Analysis
➢
Monte Carlo simulations of events
➢
Reprocessing exercises
➢ Provides Storage for:
➢
Physics Data
➢
From recorded collisions
➢
From simulated collisions
➢
User Analysis Results
➢
UTA
➢
UTA_SWT2
➢
Dedicated to simulations
➢
2360 job slots; 420 TB of storage
➢
SWT2_CPB
➢
Performs all tier 2 processing jobs
➢
10,000 job slots; 5.5 PB of storage
➢ OU
➢
OU_OCHEP
➢
Performs all tier 2 processing jobs
➢
844 job slots
➢
OU_OSCER
➢
Campus cluster provides additional simulation resources
➢
2000 job slots, 700TB Storage
This chart show the cumulative contribution of CPU-seconds delivered by Tier 2's for successful processing jobs for the year (1/1 through 10/31) SWT2 is second largest in ATLAS!
➢ Supports all ATLAS activities ➢ Current Deployed Resources:
➢
20 Gbps external network connection
➢
451 compute nodes
➢
4928 cores / 9856 job slots
➢
At least 2 GB RAM per job slot
➢
32 storage servers
➢
5.5 PB usable storage
➢
1 NFS server
➢
1 admin server
➢
14 grid/service nodes
➢ Compute Element
➢
GLOBUS based access to local batch system for submitting work
➢ Storage Element
➢
Access to storage using GridFTP servers (2)
➢ Root Door
➢
Alternative access to storage
➢ Squid caching services
➢
CVMFS (distributed file system for experiment code base)
➢
Calibration data
➢ Monitoring services
➢ Dell R630
➢
CPU: 2 x Xeon E5-2640 v4 2.4 GHz
➢
Broadwell 10 cores, 20 threads / CPU
➢
RAM: 128 GB , 2.133 GHz, DDR4
➢
Storage: 2 x 800 GB SSD (RAID-0)
➢
Network:
➢
2 x 10 Gbps (SFP+)
➢
2 x 1 Gbps (Ethernet)
➢ MD 3460 Shelf
➢
60 x 8 GB SATA disk
➢
(4 x RAID 6 Disk Groups => 4 x 104 TB exported SAS drives)
➢
4 x 12 Gbps SAS connections
➢ Dell R730
➢
CPU: 2 x Xeon E5-2680 v4 2.8 GHz
➢
Broadwell 14 cores, 28 threads / CPU
➢
RAM: 256 GB, 2.133 GHz, DDR4
➢
Network: 4 x 10 Gbps (SFP+), 2 x 1 Gbps (Ethernet)
➢ XrootD
➢
Federates storage into a common namespace
➢ Core: 2 x Dell S6000 (LAG)
➢
32 x 40 Gbps (QFSP+)
➢ TOR: Dell N4032F
➢
24 x 10 Gbps (SFP+)
➢
2 x 40 Gbps (QFSP+)
➢ Legacy: Dell N2048
➢
48 x 1 Gbps (Ethernet)
➢
2 x 10 Gbps (SFP+)
➢ The Southwest Tier 2 continues to be a valuable resource for
➢ We are helping to enable the discovery of new physics at the
➢ Our contributions will grow as we add more resources ➢ For more information visit www.atlas-swt2.org