Facilities, Equipment and Other Resources
Text for use in proposals. Original version and more info is here:
https://docs.google.com/document/d/1fcc9LwYzwC8_GF-q0Ad_HSdHJweauC_UOw40SDlBBdY/edit?tab=t.0
IT Infrastructure:
Information and technical infrastructure we utilize will be managed in coordination with the General Research Information Technology (GRIT) group at UCSB. GRIT consists of a team of IT professionals providing technical support for several Organized Research Units at UCSB, including the Marine Science Institute, the home institution for the proposed effort. GRIT support covers the full range of research computing needs, including High Performance Computing (HPC), data storage services, web services and desktop support. By using open source software, commodity hardware, and investing in staff expertise, GRIT is able to minimize the cost for research IT while providing compute resources that are secure, agile, robust and cutting edge. As a team with research backgrounds, GRIT meets researcher-specific computing and storage needs by creating compute environments on top of stable, reliable, and shared hardware. This system makes it possible to serve a wide variety of projects while maintaining high efficiency. HPC resources are available to researchers in a wide variety of standard and custom configurations with single systems up to 2 TB RAM and 128 Cores for data processing and total resources of more than 25TB of RAM, 250,000 CPU and 20+ PB of storage. GRIT staff are available to assist users in gaining access to these and other larger on and off campus computing resources, including supercomputing resources available via external providers.
Other facts which could be included above:
2024-DataSHOREs version: much more UCSB oriented.
Facilities and Equipment
The University of California, Santa Barbara (UCSB) offers extensive cyberinfrastructure resources to support data-intensive research across disciplines. These facilities will be leveraged to implement and support the proposed DataSHORE system.
North Hall Data Center (NHDC)
The NHDC serves as UCSB's primary data center, providing a robust environment for hosting critical research computing infrastructure. Key features include:
• 4,500 square feet of raised floor space
• 105 seismically-isolated 42U racks in hot/cold aisle configuration
• 200 tons of cooling capacity
• Comprehensive environmental monitoring (power, fire, smoke, water, HVAC)
• Video surveillance and alarm-based security
The NHDC connects to CalREN2 via a 100 Gbps backbone, with 10 Gbps and 40 Gbps connections available to racks and buildings across campus. This high-speed network fabric will enable rapid data transfer to and from the proposed DataSHORE system.
Elings Hall Data Center and the California NanoSystems Institute (CNSI)
Complementing the NHDC, the Elings Hall/CNSI facility provides additional compute density:
• 1,300 square feet with capacity for 40 racks
• Underfloor cooling plenum with hot/cold aisle containment
• 65 tons of cooling capacity
High-Performance Computing Resources
UCSB maintains several HPC clusters available to campus researchers:
1. Campus-wide cluster: ~4,000 cores, multiple GPUs, high-speed interconnect, parallel filesystem
2. Condo clusters: 3,500 cores, 40 GPUs, InfiniBand interconnect, parallel filesystem
These existing HPC resources will complement the DataSHORE system, allowing researchers to perform compute-intensive tasks on data stored in DataSHORE.
Center for Scientific Computing (CSC)
The CSC provides critical expertise in scientific computing, supporting researchers across disciplines. Their staff will play a key role in integrating DataSHORE with existing campus cyberinfrastructure and supporting users. The CSC maintains:
• 1,500 TB of archival storage
• Experimental systems for GPU computing development
• Licensed scientific software (Intel compilers, Lumerical FDTD, Ansys, Mathematica, MATLAB)
Marine Science Institute (MSI) Server Room
The MSI server room is 250 square feet, has 5 server racks and cooling that leverages the campus chilled water loop. With the minimal investment of 100Gb network infrastructure this space will provide a robust and sustainable facility which also distributes the DataSHORE infrastructure across campus with room for expansion.
Additional Research Computing Resources
Several research groups and institutes maintain specialized computing resources, including:
• "Fat node" scientific virtual machines totaling hundreds of cores
• Over 10 TB of RAM
• 20+ GPUs
• ~250 TB of high-speed storage integrated with research software
.