Status of UTA IAC RAC

1
Status of UTA IAC RAC

• Jae Yu
• 3rd DØSAR Workshop
• Apr. 7 9, 2004
• Louisiana Tech. University

2
UTA IAC MC Farm and Analysis Resources

• Have two independent farms
• Swift farm (HEP)
• 22 dual P3 900MHz
• 250Mbyte/cpu
• A total of .6TB disk space
• Generated a total of 3.6M events w/ mcp14
• CSE Farm
• 10 P3 900MHz cpus w/ 128M memory
• Retired ? Split into two separate farms for CSE
  Grid computing project
• Generated 1M events w/ mcp14
• Analysis Machines
• Four desktop machines of 2GHz 0.5GB each
• Storage space 0.4TB

3
The UTA-DØ Grid Team

• Weve been working well with CSE colleagues
• Faculty 2 HEP 1 CSE
• Senior Research Associate 0.1 FTE
• Research Associate 1 FTE at FNAL
• Software Program Consultant ½ FTE
• Engineering MS Students 1
• Physics Undergraduate Student 2

4
UTA Software Products

• McFarm Job control (Drew Meyer)
• As of this workshop, McFarmDSTs re-processed
  output are certified
• Ganglia resource monitoring (D. Jenkins)
• Contains 7 farms (gt350 CPUs), including Tata
  institute, India
• McFarmGraph Simulation production job monitoring
  system using gridftp (Aditya Nishanda, Sankalp
  Jain)
• Provides detailed information for a MC request
• McPerM MC Farm Performance Monitoring
• Langston took this over
• Job Manager McFarm (JMM) for SAM/Grid interface
  (HyunWoo Kim)

5
UTA RAC

• 84 P4 Xeon 2.4GHz CPU 202 GHz
• 7.5TB of Disk space

• 100 P4 Xeon 2.6GHz CPU 260 GHz
• 64TB of Disk space

• Total CPU 462 GHz
• Total disk 73TB
• Total Memory 168Gbyte
• Network bandwidth 68Gb/sec

6
UTA RAC

• Joint HEP and CSE project
• HEP DØ Regional Analysis Center
• ATLAS DC Production
• Grid 2003 test bed
• Video streaming
• Database research
• Skin resin image database
• Brought online at the end of October, 2003
• DØ MC Production begun at the end of Nov. 2003
• Generated and delivered over 7M events ? 1.75M/Mo
  (Double the DØ capacity in 2003)

7
Future Projects

• Through the end of May Participate in TMBfix 2
• 20 of the total will be fixed at DPCC
• First step toward the analysis at UTA RAC
• The fixed TMB will be 100 stored at UTA
• McFarm (Meyer Jenkins)
• Incorporate TMB fix
• Test RAW reprocessing in the summer
• Improve performances
• Develop utilities

8
Future Projects

• Inclusion of JMM into SAM/Grid regular release
• Installation of calibration database proxy at UTA
• Participate in DØSAR grid MC production
• MySQL McFarm Monitoring Database for both
  McFramGraph and McPerM (Jenkins)
• Automated MC Request download and production
• Storage and refreshment of TMB data at RAC
• Prepare for regional analysis

9
Conclusions

• DØSAR has been the driving force for the great
  progress last 6 mo
• McFarm improvement is in its mature state
• Ready to participate in any DØ activities
• SAM/Grid and McFarm are interfaced
• Official request is being processed
• Some analyses have been performed at UTA IAC
  RAC
• UTA RAC has been online for the past four
  months ? Immediately doubled the entire DØ 2003
  capacity
• We are ready for grid based MC production
• First proposal (PHANTOM) as the collaborative
  effort went in ? Must try for more