Belle and ATLAS Experiments
Project Leader: Glenn Moloney, University of Melbourne
Description
This project is deploying a common data grid infrastructure to enhance the international competitiveness of Australian physicists participating in two international frontier experiments in High Energy Physics: the Belle experiment at the KEK laboratory in Tsukuba, Japan; and the ATLAS experiment at the CERN laboratory near Geneva, Switzerland.
Belle is a running experiment producing approximately 30 Terabytes of data per year. A federation of SRB databases has been established for distribution of Belle data between the KEK laboratory and collaborating institutes. A Belle SRB database has been established at the APAC National Facility and federated with the central data store at the KEK laboratory. This project is providing the grid computing services layer provide Australian physicists with real-time access to the Belle data, and the means to participate in mass, distributed data analysis and simulation.
This project is enabling full Australian participation in the world's largest production data grid service: the Worldwide LHC Computing Grid (WLCG), operating as a deployment of the EU's EGEE project. The WLCG infrastructure (LCG2/gLite) is being deployed on computing and data centres in Australia for processing ATLAS data and support implementation tests and important physics simulations leading up to the commencement of data gathering at the Large Hadron Collider (LHC) in 2006. The ATLAS experiment will produce over four petabytes of data per year.
Achievements
| 2004 |
- With leadership from the Australian group, a Belle Data Grid consortium has been developed, with the establishment of an SRB federation between institutes in Australia (National Facility), Poland, Japan, China, Korea and Taiwan. This is the most distributed SRB federaton in operation.
- Generated over 1 Terabyte of simulation data for the use of Belle researchers utilising compute resources from VPAC, APAC, ac3 and University of Melbourne. SRB is used to distribute input data from the KEK laboratory and to make the output data available to the entire collaboration.
- ATLAS data challenge 2 was successfully completed by the Advanced Research Computing Centre at the University of Melbourne, participating as a node of the NorduGrid.
- Deployed an LCG data grid node at the University of Melbourne, and commencing the installation of an LCG node at VPAC and the Advanced Research Computing centre at the University of Melbourne
|
| 1H2005 |
- LCG2 middleware has been deployed and tested using VPAC resources.
- A meta-scheduler has been developed to production status and used to deploy Belle jobs to an LCG2 resource at VPAC.
- SRB has been successfully deployed between APAC and KEK to support the Belle activity.
- The LCG2 middleware suite has been successfully deployed on APAC resources as an initial trial Grid resource. Centrally maintained components were installed at The University of Melbourne and the Compute and Storage Element (CE, SE) were implemented on resources at VPAC and the Melbourne Advanced Research Computing Centre.
- Belle Monte Carlo jobs have been successfully tested on this LCG2 installation. This involved the translation of existing job submission procedures to work with our meta-scheduler (GQSched) and Grid middleware (LCG2, Globus Toolkit, SRB).
- A fully automated run procedure has been developed to aid in future submission of Belle Monte Carlo jobs to the expected LCG2 infrastructure. This includes automated data discovery, data preparation, data and software environment validation, analysis of output logs, and reporting of job failure or completion.
|
| 2H2005 |
- The NG-LCG gateway virtual machine has been developed and deployed at the VPAC site on the National Grid. The LCG services have also been deployed at the Advanced Research Computing centre at the University of Melbourne.
- Initial deployment of an Australian Tier 2 service at the University of Melbourne, including gLite VOMS, VOMRS, SE, LFC, CE and BDII. This is the result of an extensive collaboration with the Asia Pacific Regional Tier 1 centre at Academia Sinica, Taipei, and the WLCG operations team based at the CERN laboratory near Geneva.
- In November 2005, Marco La Rosa and Glenn Moloney co-hosted an LCG deployment workshop for the Belle collaboration at the KEK laboratory in Tsukuba, Japan. Dr. La Rosa led the tutorial workshop which deployed a working grid of LCG for Belle collaborators. Establishment of a Belle VO hosted at the KEK laboratory.
- Initial Belle Analysis Data Grid application layer has been deployed in Australia. The KEK laboratory is replacing and upgrading their entire computer system, which has delayed the deployment of the Belle Analysis Data Grid application layer to the rest of the Belle collaboration. It is expected that this may proceed before June 2006.
- The first test deployment of the Belle Monte Carlo production on the National Grid, utilising ARC at the University of Melbourne and VPAC machines brecca and edda. This includes production status of the GQSched resource broker/scheduler for brokering data and computer requests between the SRB federation and the APAC NG computing resources.
|
Plan for 2006
Through to the end of 2006, this project will move the Belle grid project through to a production level service for Australian High Energy Physicists. The services deployed and operated with support of this project will provide the primary data anlysis infrastructure for physicists at the Universities of Melbourne and Sydney. We will also see the deployment of the data and computing services to the wider Belle collaboration, with a particular focus on collaborators in Poland, US, Taiwan, Korea and Japan.
In 2006 we will see the full deployment and integration of the first Australian sites of the Worldwide LHC Computing Grid. The prototype Tier 2 site will be established as an integrated node of WLCG, participating in the WLCG Pilot Service operations commencing in June 2006. The Australian Tier 2 site will participate in the first Tier 1 – Tier 2 operations tests for the ATLAS experiment in late 2006.
Milestones for 2006
| June |
- Belle data analysis performed on a production international data grid.
- Searchable metadata databases for Belle data.
- Begin processing data from the ATLAS experiment.
- Validate processing of ATLAS data in Australia.
|
| December |
- Complete ATLAS Data Challenge 3 (atlas-DC 3) and WLCG Service Challenge 4 (SC4).
|
Participating Organisations
- University of Sydney
- Australian National University
- University of Melbourne
