Universities Migrate Research Computing to the Cloud

Researchers Need the Right Tools for Their Jobs

Jenay Robert, senior researcher at EDUCAUSE, says the cloud has become essential for supporting research as institutions recognize the importance of unifying data models. She notes that effective communication between IT departments and researchers helps smooth the path to the public cloud. “That’s one of the biggest challenges: making sure everyone is aligned so they can move through these processes in a timely way that protects data without impeding research,” she says.

As Bentley began deploying Azure resources to faculty members, the IT department produced different user personas to guide resource allocations. Initially, Shah says, the team planned to design most individual cloud environments around “Jimmy,” a user with needs for multiple CPUs or GPUs and high-speed storage for large data sets. However, they soon found that the bulk of researchers are more like “Bob,” a user who simply needs a fast machine and adequate storage capacity.

Instead of deploying identical cloud environments by default, Shah’s team surveys faculty members to learn more about the resources they need for their specific research.

“Each of these environments come with different costs,” Shah notes. “We compare them to cars: If you’re going to the grocery store, are you going to take out your Ferrari? We wanted to build for the actual need.”

RELATED: Find the technology solutions to take your university’s research to the next level.

The Cloud Offers Instant Scalability for Research Teams

In March 2023, Google and the University of California, Riverside announced a first-of-its-kind partnership that allows the school to access Google’s cloud computing resources at a fixed subscription rate.

“The cloud, for us, solves a unique problem,” says Matthew Gunkel, UC Riverside’s CIO. “We don’t have a lot of data centers. We don’t have land or the money to build new data centers. We needed a strategy that would allow us to scale within the confines of our available resources.”

At the time of the move, Gunkel predicted that it would double or even triple the school’s computing and storage capacity, and he says that the partnership has yielded immediate benefits.

“Having flat-rate access to cloud computing and open availability of the entire service catalog has been transformative in helping onboard researchers,” he says. “People are maturing through the process, from the lift-and-shift of workloads to using cloud services for building efficiency around how they conduct their research.”

Gunkel says his team is still on a learning curve, figuring out things like automated controls, management and how to prioritize requests. Still, he says, professors are already speeding up their work due to the availability of on-demand, scalable infrastructure.

“Before, researchers had to wait for things to be installed and configured,” he says. “Today, people reach out to tell us what they’re trying to achieve and we’re able to spin up resources in the cloud.”

“Shrinking that turnaround time has been significant,” Gunkel adds. “We’ve seen researchers apply for additional grants that they previously wouldn’t have been eligible for, because now they have the infrastructure they need to do the work.”

Finding the Best Fit for Research Computing

Rob Fatland, research computing director at the University of Washington, likens his school’s research computing environment to the Wild West.

“Everybody gets to choose their own path,” he says. “The upside of that is the freedom it affords faculty. The downside is the prospect of missing out on opportunities or choosing a less efficient method.”

Fatland estimates that less than 30 percent of UW’s research currently incorporates public cloud resources (mostly from Amazon Web Services, Google and Microsoft Azure).

“Our best customers are people who have reached a point where they really need to figure out how to obtain the computing resources to get their research going,” he says. When these researchers reach out, Fatland listens to their needs and then walks them through their options, including public cloud providers.

“I talk to them about what it is going to cost, not just in dollars but also in terms of time,” he says. “It’s like being given a puppy: It’s a responsibility to spend some time learning the ropes.”

Researchers who invest that time typically see a worthwhile return, Fatland says. The greatest cloud use, he says, comes from researchers in scientific and engineering departments who deal with large amounts of data, as well as those in fields with sensitive data, like medicine, who are attracted to the security guarantees made by large cloud vendors.

Fatland says he aims to grow awareness of the cloud, even if on-premises computing is a better fit for some researchers.

“There are a lot of other options available, and that’s fine,” he says. “My main goal is to allow people to forget about their computing platform and just get on with their research.”