Open OnDemand helps University of Otago expand access to HPC across campus and the Pacific
When the University of Otago –Whakaihu Waka set out to modernize access to its high performance computing (HPC) resources, the goal wasn’t to grow infrastructure — it was to remove barriers.
Located in Dunedin, New Zealand, the University of Otago serves more than 21,000 students and operates five main campuses and centers across the country, including sites in Christchurch, Wellington, Auckland, and Invercargill. While the university issues staff and researchers standard laptops suitable for most day-to-day work, those devices often fall short for computational research.
Historically, computational resources were siloed within individual departments or labs and accessing them often required command-line proficiency and familiarity with complex workflows, a barrier that left many researchers underserved. Those with significant computing needs could turn to national facilities, but the barrier to entry there was often higher, and those systems were less suited to smaller-scale projects. A more accessible, centralized approach was needed.
To address the HPC access gap, Otago turned to Open OnDemand — a web-based platform developed by the Ohio Supercomputer Center (OSC) that provides secure, browser-based access to HPC systems.
Centralizing access to advanced computing
“Open OnDemand shifted how we think about HPC infrastructure,” said Peter Higbee, eResearch solutions engineer at Otago. “Instead of treating it as a specialist resource that requires extensive training, we could position it as a collaborative research tool. That change in framing, from exclusive to inclusive, has cascading effects on who feels empowered to use it and what kinds of work become possible.”
Recognizing that opportunity, Otago pivoted. The university invested in a centralized Slurm cluster and research storage environment, with Open OnDemand as the primary access point. Instead of maintaining underutilized lab machines and complex research virtual machines, the university consolidated its computational resources. The centralized environment includes 46 cluster nodes with over 6,000 CPU cores, GPU resources, and 5 petabytes of high-performance storage.
Through a secure web browser using single sign-on and two-factor authentication, Open OnDemand enables users to launch interactive desktops, development environments, and graphical applications. More than 30 applications are available, including RStudio, JupyterLab, and containerized GPU-enabled tools. Integration with central research storage and Globus supports secure data transfer and cross-team collaboration.
“Open OnDemand enabled us to centralize our HPC infrastructure more effectively,” Higbee said. "By consolidating access through a single platform, we can manage resources more efficiently, monitor power usage across the cluster, and offer researchers access to more powerful hardware without the fragmentation that comes with scattered access methods. It’s made our infrastructure both more capable and easier to operate.”
The cultural shift was measurable.
Before Open OnDemand, HPC users were in departmental or lab silos all funded from separate grants. There was duplication of effort and resources, with inconsistent levels of support, inequitable access to compute facilities, and no economies of scale.
“The real win was accessibility,” Higbee said. “Open OnDemand means we can invite researchers from non-computational backgrounds into complex HPC workflows without gatekeeping resources behind terminal skills. It created space for genuine collaboration instead of forcing everyone through the same technical bottleneck.”
By centralizing research home directories, application environments, and storage, Open OnDemand enabled cross-lab and cross-department collaboration that previously wasn’t possible. Departments that historically did not rely on HPC — including economics, media, and humanities — are now exploring computational approaches that were once out of reach.
“My biggest piece of advice is to centralize the cost of computational resources,” Higbee said. “This provides true equity across the university.”
As demand for GPU-intensive and data-driven research accelerates globally, Otago’s model has broadened access to high-end infrastructure. Researchers can now participate regardless of departmental budgets.
Expanding access across research and teaching
Open OnDemand has also reshaped Otago’s teaching and learning approaches.
Integrated into coursework across information science, computer science, and bioinformatics, the platform removes the installation hurdle that often slows students at the outset. Because it runs entirely in a web browser, students no longer troubleshoot operating system differences. They begin working in a consistent, centrally managed environment from day one.
For instructors, that consistency simplifies teaching and assessment. For students, it lowers the barrier to entry.
Through participation in the Summer Internship for Indigenous Peoples in Genomics (SING), in partnership with the University of Guam and collaborators across Micronesia, Otago enabled students to access its HPC resources remotely for bioinformatics and population genetics training.
“It enabled participants to engage with sophisticated genomic workflows immediately, without requiring command-line proficiency,” Higbee said. “It created a collaborative environment that would have been impractical to maintain through traditional secure shell (SSH) access.”
That flexibility was especially important given Otago’s infrastructure constraints. The university does not permit SSH connections from off campus, and VPN access is limited to longer-term users. Integration with Entra ID two-factor authentication ensured both global accessibility and security.
Participants — including students on Pacific islands thousands of kilometers away — were able to work in a unified computing environment, regardless of local infrastructure.
By reframing HPC from a specialist system to a shared institutional resource, the University of Otago expanded both who participates in advanced computing and what becomes possible.
What began as an effort to modernize infrastructure has evolved into a shift in who feels empowered to use advanced computing — and how far its impact can reach.
Written by Lexi Biasi