Introduction
Fundamental to supporting research in Canada are a handful of questions: How do we know what services researchers need? How do we best provide them? What is the role of a central office? What are the roles of provincial organizations and institutions? How should decisions get made?
This document contains one proposed set of answers to these questions for the Compute Canada national platform; a sketch of one possible alternate future for providing computational research support in Canada. Its intention is to begin a discussion within the community, where we look forward together and ask how computing- and data-powered research should be supported.
The Compute Canada national project for supporting research through computing and data was assembled over the past ten years to drastically improve the capabilities of Canadian research. Investigators were too often limited in the scholarship they could perform because of lack of local research computing expertise, or availability of storage or computational resources. The solution to this problem was clear then, as it is now. The scarcest and most valuable of our resources, the expertise of research computing staff, grows rather than being diminished by applying it to many diverse problems; economies of scale apply to the sharing of large computational and storage resources. Thus, the best way to support Canadian research is to provide and share resources, rather than building small silos.
No one can deny that the research that has been enabled by the hardware and expertise provided over the years is, and continues to be, exceptional. However, like any organization it must strive to be better and challenges must be addressed with input from the entire community. Over time, the project named Compute Canada has gone from being a loose association of occasionally-cooperating independent sites to an organization with a sizeable central office where a great deal of decision making is consolidated. In this document, we argue that neither approach is sustainable; neither provides the best results for Canadian research; neither is ambitious enough in what we can do for our researchers and scholars. We instead present a vision for Compute Canada for the next ten years that is national, but not centralized; that is diverse, but interoperable; and that is focussed on supporting Canadian research with a wide range of services reflecting the breadth and depth of Canadian scholarship.
Computation and data plays a central role in all fields of study; the national project should aim for nothing less than to give Canadian researchers an unfair advantage in tackling problems that are important to Canada. Suggestions have been made by Canada’s Fundamental Science Review (the Naylor report), and will be made by an upcoming Leadership Council for Digital Research Infrastructure report; but now is the time for we the community including researchers, staff, the members and all who support Canadian research to take a step back and have a real discussion about where to go to next. It is time for us to decide what we want computational support for Canadian research to look like, and how it should work.
Any Compute Canada, present or future, must be judged against a set of principles to guide the research support organization. We propose seven such principles: that a Canadian computation and data research support platform should serve Canadian research in a researcher-centred, service-oriented, and national way; and it should operate as a true federation, interoperable but not identical, collaborative, and up-to-date. From those principles, and based on the experiences of other federated organizations in Canada and abroad, we suggest concrete organizational improvements that could help us move towards our goals.
For motivation and concreteness, we begin with an attempt to illustrate what such a future federation would look like, from the most important point of view — that of the researcher.