Set across 10 campuses on four Hawaiian islands, the University of Hawaii System presents unique challenges and opportunities for providing research computing support. FY 2024 was a record year for external funding, with the system bringing in more than $615 million. CaRCC spoke with Sean Cleveland, Interim Director for Cyberinfrastructure at the University of Hawaii System, to learn how his team delivers high-performance computing and research software engineering across this distributed environment. The following Q&A has been edited for brevity and clarity.
How did IT Services Cyberinfrastructure get started at University of Hawaii?
Cyberinfrastructure was started in 2013 when Dr. Gwen Jacobs was hired as the director of the program. She brought on Dr. Ron Merrill to help with setting up a central high-performance computing cluster. The University of Hawaii was also part of the NSF-funded ACI-REF program, a consortium of universities that provided facilitation skills to help establish the program. The group was created to develop a centralized supercomputing cluster that would serve all ten campuses across four different islands.
What kinds of services and support does IT Services Cyberinfrastructure provide?
We’ve evolved since 2013, but currently we provide a centralized computing cluster as a free resource for all researchers, faculty, and students across the system. We also run this as a condo resource, where researchers can purchase compute nodes to add to the cluster. When they’re not using them, those resources are available to others.
We offer facilitation services with two different onboarding sessions each month to train researchers on accessing and using our cluster computing system and storage systems. We’ve brought research storage online in recent years, offering some free storage attached to the cluster and also a pay-per-year-per-terabyte service. This is accessible via a service called OwnCloud, similar to Dropbox or Google Drive, allowing users to access data on laptops or workstations and collaborate.
We also offer research software engineering services and data science services. We have team members who help develop workflows, software, user interfaces, and science gateways. We collaborate with researchers on writing proposals and grants, providing facilities documents and help with data management plans. Some team members can serve as co-PIs or PIs on awards.
Who are your primary clients or users?
We work with a variety of researchers, but we have a current project with our EPSCoR Track 1 award called Change Hawaii: Harnessing the Data Revolution for Island Resilience that focuses on workforce and atmospheric science , geo-science and data science. We work extensively with scientists, helping them leverage our high-performance computing infrastructure and access to national infrastructure. We host part of the Jetstream 2 national cloud at UH and help researchers access these cloud resources.
We also partner with researchers doing bioinformatics and microbiome work, building out a science gateway to help run their pipelines and workflows. We’re now helping update that to add additional metadata and data management capabilities.
We typically have about 400+ active researchers using our infrastructure each year, with about 1,500-1,600 accounts created since we started. We also support classes, with 2-3 classes of about 20 students each accessing our systems each semester, ranging from undergraduate to graduate levels.
How many people do you have delivering these services?
On the high-performance computing side, we have Dr. Ron Merrill as our HPC manager and David Charlzenbach as our HPC architect. I used to work on that side and still pitch in occasionally.
We have three research software engineers with master’s degrees in computer science and backgrounds in data science, ranging from visualization to AI, as well as database and analytics experience. They work on projects like the Hawaii Climate Data Portal and the CMAIKI microbiome science gateway. We also partner with the Hawaii State Department of Health on their Overdose to Action program, helping build a data dashboard for monitoring substance use and behavioral health metrics across the state.
We have an administrative staff member who helps coordinate events and activities, another person who handles support and some administrative tasks with a background in networking, and a communications lead who helps with our website, newsletters, and outreach. We also have an education lead who runs our graduate and undergraduate training programs, along with typically one graduate student and two undergraduate students.
How do you run your student program?
For the last three years, we’ve had a cyber training program with a couple of graduate fellows who help build and run our cyber infrastructure and data science workshops. They also develop personal independent development plans for their professional goals. We give them opportunities to hear from graduates who have transitioned to academia or industry.
When students work with us directly, we meet with them weekly to outline what they’ll be working on. We use Redmine, a project management tool, to help manage their work. If they’re handling help desk duties, they help with tickets through that system. We try to give them a breadth of opportunity to learn technical skills and access different hardware or technology, plus professional development opportunities.
What other research support providers do you work with?
We work with one of the bioinformatics cores on campus on the microbiome project. We try to partner with some of the larger departments, like the Institute for Astronomy, which runs many of the large telescopes across the island and has significant data and computing needs. We also work with our medical school cancer center when possible, although there are challenges with HIPAA and restricted data.
We partner extensively with the Texas Advanced Computing Center on an NSF CSSI Scientific Software Middleware project called TAPIS, which we use to help deliver our Science Gateway infrastructure.
What does your program do especially well?
I think we’ve been delivering HPC services especially well and helping facilitate their adoption by researchers. We get a lot of positive feedback from faculty praising access to the resources and the expertise of our staff in helping them leverage our systems or software they’re not familiar with, or in accelerating their workflows.
We’re also proud of the science and data gateways we’ve built. The Hawaii Climate Data Portal has had 45,000-50,000 different visitors accessing many gigabytes of environmental science data. We continue to work directly with researchers to productionize their science so that things generate daily and monthly data products automatically.
How is your program funded?
We have two positions that are university/state funded: our HPC manager and the director position. Everyone else is funded through soft money. Some of the grants you mentioned, and other grants we have, provide funding. We get some return to overhead from Information Technology Services and the VPR office to support research.
The university and ITS subsidize our data center, power, and cooling. In addition to being PIs or co-PIs on awards with salaries written in, we also use a consulting model where researchers who need portions of an FTE for research software engineering, data science, or system administration can create agreements with us.
How do you measure the impact of your program?
We look at usage metrics for our different services and the demographics of who’s using them. For our workshops and training events, we track attendance to see if we need to modify our outreach to target departments or demographic groups that might not be aware of our services.
We track publications that cite our resources and the assistance we provide to see our impact on science. We survey our user community yearly and ask them to share their publications with us. We’re also starting to track what awards are coming into the institution from researchers who use our resources to see our impact on grant funding.
For our community gateways and resources, we track usage metrics for visitors, files accessed, and data downloaded, which helps us understand if they’re useful and what the community is most interested in.
What kinds of activities do you do for outreach and communications?
We have a faculty advisory board that we meet with every semester. We also hold a symposium or showcase of research where we invite users of our resources to share their findings through short talks or lightning talks.
This year we’re starting outreach to new graduate students at their orientation fair, and we try to connect with new faculty onboarding sessions as well. We also meet with potential faculty candidates during recruitment to let them know about our resources.
We do direct outreach to departments, giving talks to let them know about our resources and highlighting research that’s leveraging our infrastructure. All of this is in addition to the workshops and training we offer.
What advice would you give about hiring a communications person for research computing?
Having a communications person is important for bringing in new users and recognizing the researchers who partner with us. As we make updates and changes to our infrastructure, it’s helpful to have someone focused on communicating those changes because our technical staff are focused on the technical work.
A communications person can help take some of the load off technical staff by helping to update websites, documentation, or creating videos. They can also help organize and run events for outreach. Depending on team size, it might not need to be a full-time position initially, but having someone dedicated to communications can definitely help bring in more researchers and opportunities.
You can’t just build resources and expect faculty to come use them – it’s not a “Field of Dreams” situation. You have to do outreach.
How do you organize your team’s work?
We have bi-weekly team standup meetings where everyone gives updates on what they’ve been doing and what they’ll focus on next, so the entire team has visibility into priorities. We do strategic planning every 3-6-12 months to anticipate deadlines, programs, reports, and grant activities.
We use Redmine to track issues and projects. An administrative group meets weekly to review these items and prioritize. We meet with the hardware team weekly to address issues, and also with the software team weekly. We use a combination of Zoom meetings for synchronous communication and Slack for asynchronous communication.
What are your medium-term goals?
We’re looking at being able to support more PHI (Protected Health Information) and restricted data environments, and exploring how we can support more commercial cloud providers for our researchers. We have a finite data center size with power and cooling limitations, so we’ll host what we can locally but need to help researchers efficiently leverage other opportunities.
We also look at planning for when our big grants and funding will end and how we can replace them. We consider equipment needs and the latest trends in computing like AI accelerators and potentially quantum computing.
We align our plans with the university’s mission and with information technology services’ goals to prepare for upcoming initiatives and changes.
What’s a success story you’re particularly proud of?
The Hawaii Climate Data Portal stands out. The lead PI, Dr. Thomas Giambelluca, had been at UH for over 40 years and had wanted to build this environmental data portal for 10-15 years. He wasn’t able to do it until partnering with us, when we brought the research software engineering and data science expertise to make it happen.
Before this collaboration, it would take a year to generate a monthly rainfall gridded data product for the state. Now we can do it on the first day of the next month and have it available immediately. We’re bringing this capability online daily now. This is a big win that uses our computing infrastructure, research software engineering, and data science technologies to advance research and create community impact. The portal is now used by the state, commercial entities, USGS, and USDA.
What’s your elevator pitch for your program?
Our team exists to let researchers focus more readily on their research rather than on the technical aspects of managing infrastructure. We help researchers do more and scale more than they would be able to do if they were leveraging other resources or trying to do it on their own.
We also emphasize the funding success we’ve had and the amount of funding we’ve brought into the institution. Investment in our team results in additional funding that our group directly brings into the institution and state.