Mike Folk, Robert E. McGrath (NCSA)
Reference URL: http://hdf.ncsa.uiuc.edu/HDF5/
The Hierarchical Data Format (HDF) is a file format and software library for scientific data analysis and scientific data management in heterogeneous computing environments. The HDF library and format support a variety of data types and structures, large data granules, and metadata in a variety of forms. They also provide cross platform portability, fast I/O and efficient storage. First implemented in 1988 at NCSA, HDF has successfully supported scientific and engineering research in many different communities, and is supported by many commercial visualization and data analysis packages.
Because of limitations in the design of the format and software library, a new version of HDF, called HDF5, has been implemented. In collaboration with the NASA Earth Science Enterprise, the DOE Accelerated Strategic Computing Initiative (ASCI), and others the HDF group at NCSA has done a complete redesign of the format and library, with renewed goals of storing scientific data on the next generation of vastly expanded computational and storage systems.
HDF5 supports a data model that is simpler and more powerful than that of the current version of HDF. The new format greatly increases the size and number of objects stored in an HDF5 file. The format and library include enhancements designed to improve I/O performance and storage efficiency, particularly in high performance computing environments.
This presentation provides an overview of HDF5 features, file format layout and application programming interface. Further information can be found at http://hdf.ncsa.uiuc.edu/HDF5/.