A System for Long-Term Document Preservation

This paper analyzes the requirements and describes a system designed for retaining records and ensuring their legibility, interpretability, availability, and provable authenticity over long periods of time. In general, information preservation is accomplished not by any one single technique, but by avoiding all of the many possible events that might cause loss. The focus of the system is on preservation in the 10 to 100 year time span—a long enough period such that many difficult problems are known and can be addressed, but not unimaginable in terms of the longevity of computer systems and technology. The general approach focuses on eliminating single points of failure single elements whose failure would cause information loss combined with active detection and repair in the event of failure. Techniques employed include secret sharing, aggressive "preemptive" format conversion, metadata acquisition, active monitoring, and using standard Internet storage services in a novel