Date created: | Last Updated:

Category: Communication

Description: This paper shares lessons learned at the completion of a large multi- year project (2008-2018) to move to a next-generation digital preservation repository at Harvard Library. These final stages included activities running in parallel to migrate the metadata for millions of files, train fifty-five different units to use the new repository tools, and migrate the audio files from obsolete formats. Harvard Library’s digital preservation repository, the Digital Repository Service (DRS), holds over 69 million files and 285 TB of data from across the university. When it first launched in October 2000, it was one of the first digital preservation repositories at an academic institution. At the time it was designed, there were little- to-no digital preservation/digital library standards, best practices or tools to adopt. Planning for a next-generation DRS under the name the “DRS2 project” began in 2008, with the purpose of modernizing the DRS to take advantage of the latest technologies, standards, and practices and to provide curators, depositors, and preservation staff with significantly enhanced tools. The DRS2 project included four different types of migrations: infrastructure, metadata, file format, and repository users. It included such daunting tasks as re-architecting the entire repository; re-parsing all of the millions of files; changing the underlying data model, Archival Information Package (AIP) format, and all of the XML schemas; rewriting all of the repository metadata; retraining all of the repository users; and reformatting all of the audio deliverable files. A repository migration of this size and breadth was unprecedented in the digital preservation community, so there were few preexisting examples to learn from. By sharing the experience of this project, the authors hope that other institutions can benefit from the lessons learned during a repository migration of this magnitude. This paper provides an overview of all four migrations, but delves deepest into the results, challenges and lessons learned from the metadata migration. These lessons are applicable to preservation planning and interventions, future migrations, preventing metadata and content errors, and conducting very large projects in general. This paper was the runner up for the Best Long Paper award.