A Look Inside the Box

At the heart of CRC’s work is the relationship between the research community relying on advanced computing resources and the research computing consultants and IT specialists who make their research possible. Helping to further that relationship, CRC is working to build awareness of the center’s strategic roadmap that will best meet the needs of the research community over time.

“It’s opening up the box to let people peek inside,” says Adam Hobaugh, CRC’s co-director. “We want the research community to have a picture of how we respond to their needs in the face of new technology trends in the high-performance computing (HPC) industry. We ask ourselves what gaps we have, what technology will be more valuable or less valuable in the future. And we develop metrics to steer us by listening to users.”

CRC’s plan for acquiring and retiring equipment in 2022 are outlined in the graphic at right: items denoted by + on the left side show equipment being added and items denoted by - on the right side indicate equipment being retired. The upgrades touch all segments of the CRC ecosystem, including compute, storage, and data management. Funding for these improvements comes from a complement of NIH S10  and NSF MRI big instrumentation awards, and the budgets of CRC and Pitt IT. The center’s overall data storage capacity will increase by 2 petabytes (one petabyte - PB - equals 1,000 terabytes; each terabyte equals one gigabyte).  An additional 1 PB will replace existing aging lower-performance storage. Importantly, these new storage systems can scale up across all parts of Pitt’s research enterprise, providing CRC a flexible path for future expansion and upgrades.    

Among the strategic roadmap projects in “hmm … that’s new” category are expanding research data storage and management. Feedback from users showed gaps in CRC’s storage tiers as well as tools for managing the explosive growth of data.  

“Data may be power, but researchers have to be able to use it,” says Kim Wong, CRC’s co-director. “Through the new data management tool, we will empower our users with metrics about their data to help them make informed decisions about what to delete, what to archive, what to pin to the operational storage tier because the data is active research, and who to share data with.”  

Storage and data management are pressing issues for the research community, with the balances of cost, performance, capacity, and access between local and cloud storage emerging as a complex problem that require enterprise-grade tools. CRC is investing in tiered storage to enable end-to-end data lifecycle, including the much-requested service for research data backup.

The Center will be able to release more details to come in late summer.

 “A higher level of support allows us to provide a lot of services at the free level,” explains Hobaugh. “It’s equitable to the broader community, including users in humanities and social sciences who can do quite a lot with a relatively small computing allocation.”

People are a crucial part of CRC’s strategic planning – the center is continually evaluating what research backgrounds and technical skills augment the value of the team to researchers. 

“The consultants on the team are the “secret sauce” that really makes the difference,” declares Hobaugh. “CRC’s resources are much more valuable with the consultants. HPC’s learning curve is very steep. Only users steeped in HPC could do this on their own.”

A Pitt IT team at the Network Operations Center supports all of CRC’s resources, handling installation, testing, maintenance and decommissioning.

“I have years of experience with computing, but even I get can lost when I’m using something unfamiliar, even if it is as simple as a new user interface,” says Kim Wong, faculty research consultant and CRC’s other co-director. “When I get lost, it might be hard to find my bearings. With CRC users, if no one is there to help, they can’t apply these great resources – and that means there is creative new work that never happens.”