Particle Physics Data analysis cluster for ATLAS LHC experiment

Project Information

Discipline
Physics (203) 
Subdiscipline
40.08 Physics 
Orientation
Research 
Abstract

This activity will study the ability to establish, configure and run as small analysis cluster for particle physics data analysis from the ATLAS experiment at the Large Hadron Collider (LHC). Such a cluster includes interactive part for data analysis visualization and batch component for larger scale throughput prior to visualization.

Intellectual Merit

Many Universities were funded to setup analysis clusters at their university. This required particle physicists to spend time an increasing amount of time system administrators instead of concentrating on data analysis. With increase maturity of cloud technologies, the cluster complexity issues can be solved a small group of people and then shared with a large community. Additionally, each of the virtualized data centers at the Universities can be federated together to allow sharing of resources and to enhance the existing collaborations.

Broader Impacts

This project seeks to develop an facility that can be implemented by other particle physics groups as a means of helping to solve their needs. Both Undergraduate and Graduate students will be encouraged to use, provide feedback and otherwise be trained within this project. Undergraduate and Graduate students are actively analyzing data from the ATLAS experiment.

Project Contact

Project Lead
Doug Benjamin (bdouglas) 
Project Manager
Doug Benjamin (bdouglas) 
Project Members
Peter Onyisi, Jakob Blomer, Val Hendrix, George Lestaris, Henrik Öhman, Crystal Riley, Ateeq Baig, David Lesny  

Resource Requirements

Hardware Systems
  • india (IBM iDataPlex at IU)
  • Not sure
 
Use of FutureGrid

Intent to investigate the use of cloud resources to emulate the behavior of particle physics analysis cluster found at Universities. Data taken from the ATLAS experiment at the LHC will be analyzed.

Scale of Use

Typically 15-30 VM's run periodically for several hours (< 12 hrs) at a time, with occasional scaling up to a few hundred VM's depending on the number of analyzers using the system

Project Timeline

Submitted
09/02/2012 - 10:25 
»