Fedora 4 in Production at Penn State ScholarSphere

Deploying Fedora 4, or Migrating from Fedora 3 to Fedora 4 is a challenge with built-in rewards. This series of articles, “Fedora 4 in Production” looks into why and how community members are working with Fedora 4 to enhance both collections and workflow at their institutions.

In this article Dan Coughlin, IT Manager, Digital Scholarship and Repository Development, Penn State Libraries, describes Fedora 4 in production at Penn State ScholarSphere.

• Why did you decide to deploy Fedora 4?

We were looking at developing new features for ScholarSphere, our Hydra based IR, and ultimately decided that if we had developed these features on Fedora 3.x then we would likely need to worry about rewriting these new features as well as an upgrade. Fedora 4 had additional enhancements to support larger files, and native support for RDF, but the practicality of getting to the latest version on our object store before updating our web application played as big a part as anything.

• What were the institutional strategic priorities that Fedora 4 capabilities helped to support?

The ability to support larger file uploads is a constant challenge and any solution that improves our abilities around that challenge is (almost by default) going to be supporting a strategic priority of supporting "big data" needs on our campus. We would like to do more with linked-data to improve discovery and access as strategic goals and the native support of RDF in Fedora 4 is beneficial to those.

• What were the challenges you faced in deploying Fedora 4?

At the time we were either the first or very early to upgrade to Fedora 4, this meant we were hitting some problems first. Another challenge was selling the idea of spending a lot of time and resources on migrating to a platform that wasn't going to immediately provide benefits to our users. Upgrading to Fedora 4 was much more about long term direction and strategy then what would benefit our users on day 1.

• How is Fedora working in production?

We are still on it, so that's a good sign! Initially we ran up against some performance problems that we hadn't seen in our test migrations, but those were solved within a couple weeks of our release.

• What deployment advice would you give to others in the community?

Test with real data as much as possible, and talk with others in the community that have already done the migration to see what they've learned. And as with any data migration, get it done right, not (necessarily) done fast.