A Quest to Upgrade from a Legacy to a Modern Open Source Repository
MetadataShow full item record
The Los Alamos National Laboratory Research Library is facing the inevitable need to upgrade software supporting online tools. Our legacy repository was developed in house and grew with our data model and environment - tailored to our needs but complex to maintain. We turned to existing open source software for a replacement to ease both software set up and maintenance, with the understanding that we would have to curate and standardize our data for the new infrastructure. We will present our entire workflow for this endeavor, documenting each step from the choice of an open source repository to the very few modifications needed to meet our deployment environment needs. DSpace is a lightweight infrastructure which meets our technical requirements and upkeep goals, while the DSpace-CRIS extension offers additional record management capabilities. We selected version 5.8 of DSpace-CRIS and elected to deploy it in Docker instances. This strategy suits our particular environment framework, as the Research Library seeks to replace separate user portals operated on separate networks with different access control. Docker instances running the same DSpace code can be configured for each user portal and deployed on different networks while minimizing development efforts and required maintenance. Additionally, docker containers offer independence from deployment systems, shifting the burden of system administration to docker container set up – ideal in an academic context where system administration is often the responsibility of other entities. In addition to front-end replacement, upgrading our repositories entailed migrating data from the database, as well as translating our legacy data models to one supported by DSpace. Extension of DSpace-supported schemas with custom fields allowed enough flexibility for backward compatibility and harmonization of the various data sets we migrated. We customized the DSpace code very lightly to accommodate the data model extension, provide users with a pleasant experience and satisfy our contractual requirements. Our aging software replacement strategy, while specific to our system and unquestionably not the only approach to follow, can certainly shed some light on potential hurdles and considerations faced by other libraries in a similar predicament.