By Helen Wright, CRA-Industry Senior Program Associate
The Computing Research Association‘s newest committee, Computing Research Association-Industry, held their first workshop at the end of March on Best Practices on Using the Cloud for Computing Research. It brought together 30 participants in a hybrid format in Washington, DC from industry, academia, and government. This workshop was based off of a very successful September 2021 roundtable event, in which over 50 members of the community attended. The goal of the workshop was to continue the momentum from the roundtable and identify best practices on using the cloud for computing research in three different areas, education, collaboration, and research.
It is clear that education is enhanced by the cloud. The workshop participants talked about how the cloud is a democratizer. It makes an experience uniform and equal among students in a classroom. During the height of the COVID-19 pandemic, the cloud allowed for education to continue via hybrid, remote, and/or onsite. It allowed for easy collaboration among students and gave access to specialized resources that may not have been available at all educational institutions. The continued adoption of the cloud in classes at the institutional level will provide better scale and sustainable support over the long term for multiple types of applications and uses (virtual desktop, VMs, cloud services). As a result, academia should continue to provide incentives for faculty to develop cloud based courses, industry should continue to provide course development grants related to cloud computing, and the government should continue to support cloud-based curriculum development.
Cutting-edge collaborative research in many domains is impossible without access to large datasets found in the cloud. These include healthcare data, large code repositories, large language repositories, scientific datasets, etc. Participants talked about how collaborations in the cloud can enhance research by performing data analysis and computations tasks, which in turn reduces redundancy. It is important to share these interactions to facilitate subsequent research and new discoveries. This is also true for collaborations around code. There are many examples of commercial, academic, and government open source code contributions that relate to cloud computing. Nvidia, for example, developed a federated learning framework and applied it with hospitals around the world. Finally, collaborative services, such as Google Scholar, GitHub, Globus, Overleaf, GradeScope, are not only convenient but allow the research community to grow. More collaborative resources should be incentivized and hopefully developed.
The cloud has forever changed the research community. Research methods will have to change, rather than having the cloud conform to current research methods. Data is being shared at a greater rate than ever before. The participants agreed that these changes should be embraced and encouraged. The public clouds should be leveraged. With the expansion of cloud computing, there will be heterogeneity of requirements and resources that different clouds will specialize in. These should be embraced, since requirements vary due to the nature of research. A single cloud will not meet the needs of all. A hybrid cloud or multiple cloud idea is promising, but given the nature of data and the great expansion that all fields are going at, it will not be enough.