How does Canada’s advanced research computing compare internationally?
Dr. Greg Newby, Chief Technology Officer
In July, I was part of a panel at the PEARC conference in New Orleans. PEARC is a new conference for research computing, which grew from the XSEDE conference series, and brought together advanced research computing users, service provider organizations, industry, and others to share results and discuss the future.
Compute Canada is our country’s only national provider of advanced research computing (ARC) and serves the entire academic spectrum. Other countries, or groups of countries, provide ARC in their own way. The panel, “National scale research computing and beyond,” sought to identify ways that other national or transnational ARC providers could learn from one another, and benefit from cooperation and information sharing.
Each organization represented on the panel serves a large and diverse user community and has a complex set of relationships with funding agencies, partners, providers, and other stakeholders. Each is different, but shares similarities too.
XSEDE, PRACE and EGI joined the panel. XSEDE is the US academic program, Extreme Science and Engineering Discovery Environment. It has structural and funding similarities to Compute Canada. EGI and PRACE are European consortia, with membership spanning multiple countries. They are separate organizations, each with a different emphasis. EGI has a vast service catalog, a broad range of service providers, and meets computational needs of academics and some business users. PRACE coordinates operation of the large academic supercomputers, which are accessible to users after a merit-based review process.
One common theme among the organizations is the need for various performance reports. Reports include almost everything imaginable, from budgetary expenditures to counts of CPU cycles and storage utilization. For all, the most important measures are of scientific productivity and outcomes. In Canada, the emphasis of major funding agencies on using the common curriculum vitae (CCV) has given us an advantage.
Because Canada’s research funding agencies require CCV for anyone applying for funding, Compute Canada was able to link researcher publication information to system allocation and usage. Compute Canada principal investigators, who are almost always faculty with CCV data, were asked to flag their publications that were enabled, or partially enabled, by Compute Canada resources or support.
The result is the best ever dataset reflecting publications of Compute Canada’s users. Over 2,300 faculty members provided information on over 75,000 publications. Of those publications, around half were enabled or partially enabled by Compute Canada resources or support. Panelists at PEARC were envious of these outcomes because, without the type of reporting mandate that CCV enacts, publication data tends to be underreported.
You can read Compute Canada’s full bibliometric report here.
One of the discoveries from this report was that disciplines that are relatively new to ARC, along with multidisciplinary research, had the highest scores. I believe this reflects the research strength of early innovators in those fields. These researchers might need to work harder to adjust their workflows to transition from desktop or lab-based systems to national ARC systems, but the benefits are larger than for fields that have been traditionally large users.
Several other themes were discussed at the panel, including the need for ongoing interaction with funding agencies and other stakeholders. All organizations reported challenges in obtaining the sustainable and predictable funding needed for long-term planning, and all face stresses due to a growing user base and increased demands.
Another theme that was evident on the panel, and from over 60 session attendees, is the strong passion and optimism within ARC organizations. Time and again, participants at the conference noted how they enjoy providing the support, training, operations, and other elements of national ARC systems and services. They are always looking ahead, to improvements that will help users be more productive.
I have found ARC professionals, in Canada and abroad, committed to envisioning bigger, faster, and more capable systems and services for the future. The benefits to researchers and to society are motivating us to turn those visions into reality.