Mesh Data Management in Large-Scale Scientific Computing

Abstract

In many research fields of numerical simulations, programs often produce a large amount of mesh data with complex structure. It is a fatal bottleneck for scientists to manage such large-scale simulation data. In allusion to typical data characteristics of multi-dimension and multi-variables numerical simulation programs, the relativity between time and space, and practical problems of data exchange and sharing, this paper brings forward a hierarchical data model to organize mesh data. We use meta-data to efficiently organize and manage mesh data, and have developed a unified interface for high-level data access, supporting scientific computing, visualization and other data analysis tools to store mesh data with a unified method. Meanwhile, we have completed a scientific data management system where users can browse and search meta-data on Web pages. In addition, users can browse and analyze remote distributed mesh data, and do spatial or value range queries to their mesh data.