Providing Authentic Long-term Archival Access to Complex Relational Data

Abstract: We discuss long-term preservation of and access to relational databases. The focus is on national archives and science data archives which have to ingest and integrate data from a broad spectrum of vendor-specific relational database management systems (RDBMS). Furthermore, we present our solution SIARD which analyzes and extracts data and data logic from almost any RDBMS. It enables, to a reasonable level of authenticity, complete detachment of databases from their vendor-specific environment. The user can add archival descriptive metadata according to a customizable schema. A SIARD database archive integrates data, data logic, technical metadata, and archival descriptive information in one archival information package, independent of any specific software and hardware, based upon plain text files and the standardized languages SQL and XML. For usage purposes, a SIARD archive can be reloaded into any current or future RDBMS which supports standard SQL. In addition, SIARD contains a client that enables 'on demand' reload of archives into a target RDBMS, and multi-user remote access for querying and browsing the data together with its technical and descriptive metadata in one graphical user interface.