Acquisition of a Data Depot Network for Wide Area, Data Intensive Collaboration

点击翻译按钮获取中文摘要

A Vanderbilt University-led collaboratory proposes to develop a Research and Education Data Depot Network (REDDnet) in response to the need for a shared, wide-area storage infrastructure where data flows can be buffered for collaborative processing, including analysis, reduction, visualization, exploration. The general "data logistics" problem arises when quantities of scientific or other data become large, at the level of 1TB or more per day. High performance networks help users move their data faster, but users routinely have nowhere to put their terabyte down to work on it or share it with someone. REDDnet will consist of a set of eight large storage nodes, strategically positioned across the nation's high performance research networks, and configured with a distributed storage management technology, called Logistical Networking (LN), expressly designed to attack major problems of data logistics. With more than 320 TB of projected capacity, it will be used by a diver set of researchers and educators with large datasets to manage and an established need for distributed collaboration. The initial group of REDDnet users includes two "petascale" particle physics experiments with approximately 5000 collaborators world-wide, (ATLAS and CMS), a nationwide community (AmericaView) mining information from satellite-based sensors, a consortium of eight universities and Oak Ridge National Laboratory doing large scale simulations of supernovas (the Terascale Supernova Initiative), and a collaborative effort between a Vanderbilt and Lawrence Berkeley National Laboratory investigating the structure of biological molecules (Image Reconstruction of Large Macromolecular Assemblies). The two main organizational collaborators in REDDnet, Vanderbilt's Advanced Computing Center for Research and Education (ACCRE) and AmericaView (AV), have educational missions and ongoing activities that are well positioned to exploit the significant potential of REDDnet for education and outreach.