Emerging from the dark ages
by Paul Kushner, Professor at the University of Toronto and Compute Canada User
I am a climate scientist who works with the latest generation of Earth Systems models that we use to understand and predict climate change. I’ve spent a good part of my career working with advanced research computing resources in the U.S. and Canada. I use these resources to capture the climate of the recent past and predict the climate of the coming decades. I want to understand the many facets of global warming and climate change; I’ve been focusing recently on how the oceans, sea ice, and snow cover are changing and will evolve, and what that means for day-to-day weather in Canada and around the world. The figure below shows an example of the cool computational research going on in my group.
To stay at the forefront of my field I typically need computing resources that require me to use a couple of hundred really fast processors on a supercomputer, 24/7, year round. In the last several years I have come to rely on the excellent advanced research computing facilities now available in Canada under the umbrella of Compute Canada.
I remember the “Dark Ages” before the establishment of the networked advanced research computing I now use. As a University researcher I was typically in the position of having to obtain funding to buy and maintain my own computers in my Department, or work informally with other collaborators to leverage our purchasing power with computer vendors. Finding and retaining technical support was a nightmare; we ended up routinely losing time and data because of the inefficiencies of this informal approach.
With networked advanced computing resources managed by Compute Canada and supported by nationally pooled research infrastructure funding, I am now able to request and access resources through an orderly peer review process, count on reliable systems that are managed by great IT professionals, and spend much more of my time running simulations and carrying out research.
This makes my University happy because doing research represents a good portion of what I’m paid to do, and represents a bottom line payoff in our system of publicly funded University based research. Canada’s advanced computing resources available to its academic community are envied around the world. For example, last year I used Compute Canada resources on the SciNet facility in Toronto to contribute a large set of simulations to an international project on Earth system modelling at a level that none of my colleagues were able to achieve. For me this was a real point of Canadian pride.
For academics like me, access to nationally distributed advanced research computing provides the most efficient infrastructure environment for computer-driven research. There’s even more of a business case to be made for pooling resources with my colleagues across Canada. Each year I can adjust my resource requests up and down without having to put capital funding to buy and install new computers — or decommission unused ones.
These facilities can also provide “incubator” resources for pilot projects from my colleagues who are just starting their professional research careers. As they gain experience with the support provided by these facilities, they can scale up their resource requirements to meet their expanding needs, but maintain continuity in their research approach.
In the past couple of years I have noticed that our resources are becoming more limited because of a lack of new infrastructure funds going into advanced research computing in Canada. Having built up a resource that is the envy of my colleagues around the world, it would be a real shame to return to the Dark Ages! It’s time to tell this great made-in-Canada story at the provincial and federal level. Let’s encourage our decision makers to renew the investments needed so we don’t lose our edge in advanced research computing.