Last week, I attended the UK e-Science Meeting 2011 meeting, held at the splendid Ron Cooke Hub at the University of York. This year’s topic was Towards the Cloud: Infrastructures, Applications, Research. I gave a presentation on The Application of Cloud Computing To Scientific Workflows: A Study Of Cost and Performance (by Berriman, Deelman, Juve, Rynge and Volcker), the topic of a previous post. The workshop consisted of keynote presentations, technical sessions, breakout workshops, poster sessions and exhibits. I will give an overview of the meeting here, and give more details in subsequent posts.
A theme that emerged from the meeting was that current model of local computing was starting fail across all disciplines, and in the future, computational analysis would have to be moved off local machines on to parallel processors, connected to data by high-speed networks. The role that can be played by the cloud in this computing model was the focus of many of presentations. Daron Green of Microsoft emphasized this topic in his keynote presentation on “e-Science: Enabling Data Intensive Research,” which included reports on Microsoft’s plans to make data input to their desktop applications link to seamlessly to the cloud.
In his keynote address on “Cloud: An Industry View,” Dave Pearson of Oracle described the history of grid and cloud computing, and emphasized the importance of virtualization in modern computing. He concluded with a summary of the challenges that face the cloud, including: provisioning server capacity; implementing policy; provisioning storage capacity; loss of visibility and context by the end-user.
I also sat in on a panel discussion on the topic of a “Data Community Capability Model Framework,” which discussed the need for a compute model for e-science, and discussed legal and economic issues; skills and training; collaborative working environments; scholarly communications; and standardization.