Petascale data architectures for portals and computing centers

Many science collaborations use data portals (also called Science Gateways) to provide discovery, search, and access/download services for the datasets used by the collaboration, or for the entire field. Astronomy, genomics, and climate science are just a few of the fields that make heavy use of data portals. Often, the datasets published in portals are large or complex enough that they require the capabilities of a supercomputer center for scientifically productive data analysis. This presentation will describe advances made in data portal architecture and in the production deployment of high-performance data services at supercomputer centers, and the ways in which these services synergistically interact. The Science DMZ model provides an architectural framework, and the Modern Research Data Portal design pattern builds on the Science DMZ to enable high-performance data portals which are well-matched to the high performance data capabilities of today’s supercomputer centers. A demonstration of these capabilities on production infrastructure will be included in the presentation.

Speakers

• Eli Dart Lawrence Berkeley National Laboratory