Microsoft Customer Story reproduced with permission.

At University of Bath, academic researchers work hard on difficult issues, from pharmaceutical development to astrophysics. While one team builds digital simulations of chemical reactions that will help reduce the time and cost of drug trials, another conducts blue-sky research to determine strategies for processing the literally cosmic volume of data that will come from future telescopes. Meanwhile, other teams may be working on climate change or public health.

Professor Nick Brook, Dean of the Faculty of Science, explains:

With Azure HPC + AI, we’re no longer constrained to whatever we can fit into a machine room. If we have the vision, research to support, and funding for it, the sky’s the limit.

It’s all big stuff. That’s why the University of Bath prioritizes big thinking and solid strategies in everything it does. Whether it’s investing in focused growth around science and technology or bringing the latest program and service innovations to its research and teaching community, University of Bath wants to use technology to connect all areas of the university in a seamless, powerful, and safe environment where researchers and students have the capacity to do their best work.

Since 2007, University of Bath has invested more than £8 million into supercomputing, deploying a high-performance computing (HPC) environment across every department with a focus on research needs for chemistry, mathematics, engineering, and physics. The university has recently looked to machine learning, AI, and the cloud to offer increased flexibility and agility to students, staff, and faculty researchers. In doing so, it aims to give researchers access to cutting-edge platforms and help them set the pace in academic research.

“Students coming to University of Bath expect to be at the cutting edge of technology when they come to learn,” says Professor Nick Brook, Dean of the Faculty of Science. “And the members of our research community expect to have the technology to deal with the changes and challenges in the research agenda.”

New research capacity

The latest advancements in HPC present a massive opportunity at the University of Bath to make important technology infrastructure connections and reach new heights in research abilities and capacity. It has become one of the first universities in the world to move almost all its existing HPC resources to the cloud by using Microsoft Azure HPC + AI to build a supercomputing environment it calls Janus.

“We have serious aspirations in terms of our research power with the technologies we have on hand,” explains Alex Butler, Chief Digital and Information Officer at University of Bath. “Azure HPC + AI gives us a great opportunity to really think about what that kind of model could look like.”

With Azure HPC + AI, researchers across the university have tapped into significantly reduced computing times and industry-leading computing options that will evolve to meet new research needs. Butler credits the seamless transition in part to the user-friendly Azure interface and straightforward migration process.

“In our everyday lives, we’re accustomed to using the digital tools and techniques that make up the way we live, and that’s not really different here,” says Butler. “We’re always thinking about how we can give our academics the confidence, skills, and capabilities to feel that they can use our technology and advance their research without having to lean on a huge IT team.”

Managing the data life cycle

To power its research workloads, University of Bath deployed Azure HPC resources on 21 virtual machine (VM) instances, using Azure Spot Virtual Machines, HB, HBv2, and HBv3 nodes and HC-series nodes, plus Fsv2, NCsv3, and NDv2 nodes. This breadth of VMs helps the university ensure that the right VMs can be spun up for the particular type of research project underway. “With Azure HPC + AI, we’re no longer constrained to whatever we can fit into a machine room,” says Professor Brook. “If we have the vision, research to support, and funding for it, the sky’s the limit.”

University of Bath relies on Azure CycleCloud to automate, configure, and manage the Janus environment, along with Azure NetApp Files, Azure Blob Storage, and Azure Disk Storage for meeting its storage demands. A mix of hot and warm storage, alongside temporary scratch storage options, gives researchers the opportunity to select the speed, cost, and capacity that works for them in the moment.

“Azure Blob Storage is one of the most critical things we’ve identified so far in this project - it holds the whole life cycle of the data,” says Dr Stefano Angioni, Senior Cloud Architect at University of Bath. “The data has to flow through the system from a cheaper storage to a more performant storage and back. Ideally, we’d like to automate this in the future, helping researchers make the best and most efficient use of their funding.”

University of Bath considers Janus a phase-one environment, designed to build up a user base within Azure. The cloud-based Janus complements Anatra, the university’s single, on-premises, high-throughput computing (HTC) cluster that will soon be decommissioned in the university’s phase two pursuit of moving fully to the cloud.

Both phases mark the culmination of University of Bath’s two years of exploration into the sustainability of cloud infrastructure. This involved an HPC cloud service pilot project in 2020 supported by implementation partner Red Oak Consulting that proved HPC’s value in computational fluid dynamics simulations - a process that has become one of Janus’s most common workloads.

More features, faster turnaround, and critical data safeguards

Establishing its Azure HPC + AI environment helps the University of Bath rapidly accelerate research outcomes, leading to exponential progress in resolving world issues. According to the university, it is now feasible for three years’ work to be done in three months, making speed a major selling feature for encouraging more academics to realize the potential of Azure HPC + AI for their research. The IT team can offer researchers more features with faster turnarounds while still safeguarding sensitive data, and it can be confident that research teams always have access to the latest, most advanced tooling - all tried and tested by Microsoft.

Dr Stefano Angioni, Senior Cloud Architect, explains:

Operationally, we can benefit from all these great services in Azure and take advantage of them to deploy clusters in a couple of hours.

The benefits don’t end at research outputs. Several University of Bath student courses, including a scientific computing course in mathematics and parallel computing course in computer science, have fully migrated to the cloud. “I think students recognize that they will need to be more programming literate and more aware of the computing environment that will face them in the future world of work,” says Professor Brook. “And they want to learn those skills now in a safe environment.”

A broader digital transformation

Given all its progress, the university has decided to make its broader digital transformation - planned to span the next five years - almost fully cloud based. “Cloud computing will underpin a lot of the work we’re doing, most of which is centred around not just migrating applications to the cloud but actually rethinking the interaction of our systems, applications, and data,” says Butler.

The university is already starting to explore its community’s interest in Azure Cognitive Services, better uses of AI, and how to work with big data in new ways thanks to the increased capacity offered by Azure. “I think we’re now at the forefront of cloud and HPC,” says Professor Brook. “We made what I think is a brave decision that others hadn’t taken - and we’re seeing the clear benefits.”

Alex Butler, Chief Digital and Information Officer, stated:

We have serious aspirations in terms of our research power with the technologies we have on hand. Azure HPC + AI gives us a great opportunity to really think about what that kind of model could look like.