Enabling Computer and Information Science and Engineering Research and Education in the Cloud Workshop

Princeton University, Princeton NJ

Investigators

Abstract

-Part 1: Cloud computing can transform computer science research and education through on-demand, elastic, and self-serve access to computation and storage resources at scale, coupled with contemporary hardware (e.g., graphical and tensor processing units), advanced software stacks (e.g., machine-learning libraries), and shared data sets. Researchers and educators can avoid the substantial time, energy, and expense of building and maintaining local infrastructure, students can be better prepared for the cloud-centered world they will enter upon graduation, and research projects across institutional boundaries can more easily share and maintain data and the results of their analysis. Although academic adoption of the cloud is increasing, significant challenges remain. Researchers worry about the relative costs compared to local infrastructure, especially when some local costs are hidden (e.g., "free" rack space, cooling, and power, as well as lower overhead charges on grants for equipment purchases). Graduate students worry about causing run-away costs if their experiments in the cloud run amok, and the risk of escalating costs near major paper submission deadlines. Some projects may not be able to use the public cloud, due to privacy issues (or the perception of increased privacy risk) concerning the underlying data. Plus, for both researchers and educators, making the transition to the cloud requires overcoming a "learning curve" to select a particular cloud offering---and learn how to use it effectively. This workshop will identify opportunities and challenges for Computer and Information Science and Engineering (CISE) researchers and educators to use the cloud, and recommend steps the major stakeholders can take to lower the barriers to cloud adoption. By convening a diverse community of stakeholders who can contribute to the strategic development of a cloud-computing roadmap, this workshop will lower the barriers for cloud adoption for academic research and education in CISE. The perspectives and insights shared from industry, government, and academia, coupled with discussions focused on actionable challenges and measurable outcomes, will serve to advance CISE research and education. The workshop will identify high-priority topics and high-impact opportunities for future research, training, and collaboration in conjunction with cloud computing. -Part 2 The workshop will engage about 35-45 participants from academic, industry, and government, including principal investigators, educators, cloud-computing researchers, campus CIOs, public cloud providers, and government funding agencies. The workshop will focus on several key issues including, (1) CISE research and the cloud: What types of CISE research are (or are not) best enabled by cloud platforms? (2) CISE education and the cloud: What types of CISE education are best enabled by cloud platforms? What novel concepts can be taught? (3) Key challenges with cloud usage: What are the barriers to cloud adoption and how should the stakeholders address these issues? (4) Relationship among various types of resources: What roles should on-campus resources, national resources, and cloud-provided resources each play in supporting research and teaching? (5) Cloud costs: How can researchers and educators best manage the costs of cloud usage, including the risks of runaway costs or costs that continue after a grant funding a project has ended? (6) Education for prospective cloud users: What are good ways to educate and train researchers, students, and campus IT staff on using the cloud effectively? (7) Ongoing user support: What ongoing support is needed for academic users, whether locally or at the national level? The workshop will host a series of presentations and working sessions with live note-takers, resulting in a post-meeting white paper describing key themes, insights, and recommendations from the discussions. Lowering the barriers to academic use of the cloud has the potential to significantly improve how CISE researchers and educators conduct their work. Using the cloud can reduce IT costs, foster cross-institution and interdisciplinary research collaborators, and better train students for a cloud-centric world. Innovations in cloud computing services can help cloud providers recognize new opportunities ahead of future commercial demands, while better supporting academic research and education.