A. CiocioITSD/CSAC Retreat March 3, Scientific Cluster Support Program SCS Steering Committee Report

A. CiocioITSD/CSAC Retreat March 3, Overview Growing interest in the use of Linux clusters for scientific research at Berkeley Lab Hard to efficiently manage a multi-node compute cluster Findings from the Berkeley Lab Midrange Computing Workshop (March 2002) and subsequent discussions with scientists identified a need for affordable centralized support The ultimate goal: to increase the use of scientific computing to Lab research projects to introduce parallel computing to Berkeley Lab researchers to develop efficient, cost-effective methods for managing production clusters. Four year program started 1/03

A. CiocioITSD/CSAC Retreat March 3, Mid Range Computing Gap

A. CiocioITSD/CSAC Retreat March 3, Scientific Cluster Support Program Ten research projects from seven of the Lab's scientific Divisions were selected to participate in the 4 year Laboratory-funded program after a Lab-wide application process that was completed in September These projects are eligible to receive the following services: Pre-purchase consulting Procurement assistance Setup and configuration Ongoing systems administration and cyber security Computer room space with networking and cooling

A. CiocioITSD/CSAC Retreat March 3, Chemical Sciences PI: Miller Contact: Michele Semiclassical Molecular Reaction Dynamics: Methodological Development and Application to Complex Systems New 30 Node Linux Cluster Chemical Sciences PI/Contact: M. Head-Gordon Parallel electronic structure theoryNew Node Linux Cluster Chemical Sciences PI: Lester Contact: Aspuru- Guzik Quantum Monte Carlo for electronic structureExisting 19 Node Linux Cluster Materials Sciences PI: Arup Chakraborty Contact: Hagan, Dinner Signaling and Mechanical Responses Due to Biomolecular Binding New 50 Node Linux Cluster Physical Bioscience PI/Contact: Kim/Adams/ Brenner/Holbrook Structural Genomics of a Minimal Genome Computational Structural & Functional Genomics A Structural Classification of RNA Nudix DNA Repair Enzymes from Deinococcus radiodurans Existing 32 Node IBM Linux Cluster to be expanded to 64 Nodes Environmental Energy Technologies PI: Gadgil/Brown Contact: Finlayson/Tonse Airflow and Pollutant Transport in Buildings Regional Air Quality Modeling Combustion Modeling New 20 Node Linux Cluster Earth Sciences PI: Hoversten/Majer Contact: Daley Geophysical Imaging Existing 16 Node Linux Cluster to be expanded to 24 nodes Life Sciences PI/Contact: Eisen Computational Analysis of cis-Regulatory Content of Animal Genomes Existing 6 node Linux Cluster to be expanded to ~ 20 nodes Life Sciences PI: Cooper/Tainer Contact: Frankel Protein Crystallography and SAXS data Analysis for Sibyls/SBDR New 16 Node Linux Cluster Physics PI/Contact: White Computing the Universe Existing 10 Node Linux Cluster to be expanded

A. CiocioITSD/CSAC Retreat March 3, SCSC The SCS Steering Committee is a working group of CSAC. As such, it informs CSAC of SCS project status and issues on a quarterly basis, or as needed, and solicits input from CSAC members as needed to aid in decision making and priority setting. In turn, CSAC members, as representatives of their divisions and CSAC, have the responsibility to communicate information to the Steering Committee that is important to its role in governance of the SCS Project. The SCS Steering Committee chartered by this document will work with the full CSAC committee and the ITSD project team to ensure success and visibility of the first phase (implementation phase) of the SCS Project. This phase will conclude once all of the clusters have been purchased or integrated into the project, and the support activities for the clusters have become routine, anticipated March The Steering Committee will meet bi-monthly or as-needed during this first period. The Steering Committee is responsible for governance of the implementation phase of the project. ITSD is responsible for day-to-day management. This governance includes the following activities: - Provide oversight to ensure accountability - Participate in decision making - Participate in priority setting

A. CiocioITSD/CSAC Retreat March 3, SCSC Members Committee Chair - Alessandra Ciocio - Physics, CSAC and MRC Working Group Paul D. Adams - PBD, CSAC and MRC Working Group Shane Canon - NERSC Tom Daley - ESD Damir Sudar - LSD, CSAC Gary Jung - SCS Project Manager Tammy Welcome - SCS Project Director Jim Triplett (advisor)

A. CiocioITSD/CSAC Retreat March 3, Status SCS Program progressing well despite complexity of project scheduling due to availability of each project's procurement funds and differences in readiness. 6 of 10 clusters in production (Chakraborty, Gadgill/Brown, Hoversten/Majer, Miller, Lester, Eisen) 2 clusters in progress (Adams/Kim/Holbrook/Brenner, Cooper/Tainer) 1 cluster upcoming (Head-Gordon) 1 cluster opting out (White)

A. CiocioITSD/CSAC Retreat March 3, Progress Development of Linux cluster expertise has encouraged the use of Linux clusters at Berkeley Lab. These are funded on recharge and include: PBD Berkeley Structural Genomics Center (14 processors) LANL Tuberculosis Structural Genomics Consortium (24 processors) Yucca Mountain Project (12 processors) Yucca Mountain Project (32 processors)

A. CiocioITSD/CSAC Retreat March 3, Developments ITSD has developed high quality cost-effective Linux cluster support for Berkeley Lab - Development of standard toolsets and procedures Warewulf Cluster Implementation Toolkit Developed locally by Greg Kurtzer GPL licensed through Lab Tech Transfer Showcased at Supercomputing 2003 Used widely outside the Lab (Univ of Kentucky supercomputer) - Higher level of cybersecurity (SecureID cards) - Continual review of support costs Summary: 360 processors in production (278 SCS, 82 Non-SCS)

A. CiocioITSD/CSAC Retreat March 3, Past and present Issues for the SCS Steering Committee Conforming to developed standards (cluster distribution, architecture) US Export Controls (deemed export) How do we select SCS replacements or additions? - What to do beyond SCS program? - Extensibility - Many other projects are interested - Revisit institutional computing resource - What about GRIDs?