| Abstract: | Computational and laboratory experiments generate masses of data that must be stored reliably, with minimal effort on each researcher's part, and must be retrievable for decades. The storage environment must also work seamlessly across scientific disciplines and capture all of a file system's features in a semantically-based catalog that provides Boolean, keyword, and tree-based data access. The authors describe a metadata-based archive for scientific data that provides flexible archive storage for very large data sets. The system uses metadata to organize and manage the data without imposing predefined metadata formats on scientists. |
