Fedora/Hydra (and DSpace)
Breakout Session Notes from CurateCAMP SE (Atlanta 2012)
- Review of Hydra, Islandora and other Fedora applications
- Fedora security implementations:
- FeSL implementation at UNC
- Ruby-based security at Indiana
- storing permissions in SOLR for faster access
- fedoraAdmin account vs. real users in the audit trail
- Ingest considerations
- admin client
- REST API vs. SOAP API
- uses of Fedora java message service
- storage implementations
- EMC's Akubra module
- iRODS Akurbra module
- hierarchal storage and storage hints
Conceptualizing the Discussion
- Triple-store Solutions
- Triple-store Solutions
- What options are there for authentication and access control?
- What are some of the benefits and pitfalls of particular approaches?
- How will Fedora behave in a linked-data world? What capabilities will it provide?
- Can something other than Fedora (ie. Mulgara) be used for triple-store metadata?
- What is the extent of transaction management within Fedora?
- How does Fedora handle large (30-40GB) data streams?
- For linked data, we must structure the Fedora architecture in such a way as to provide a framework for linked data
- What specifically does this entail?
- Triple-store metadata can be setup to something like Mulgara but it must be setup manually (through something like Trippi, a layer between triple-store and Fedora)
- JMS (Java messaging service) can be used for transaction management, but it is not quite atomic.
- There is something called "Fedora Management Data Streams" for handling large data streams, accessible via REST API.
Fedora Related Software
- Solr tool for searching, indexing, faceting.
- Being used to manage multiple Voyager implementations at Wisconsin University schools.
- A Drupal based front-end for Fedora.
- Collection level editing
- Some built-in workflows
- Compatible with Drupal 5 & Drupal 6 (not Drupal 7).
- Fedora Enhanced Security Layer (FESL)
- Storage Abstraction Layer for Fedora
- Can utilize IRODS, S3, NAS, etc
- Can enable "hinting" for destinations
- Fedora Enhanced Content Models
- Enables some content validation abilities
- Active Fedora
- Opinionated Metadata (OM)
- Nokogiri - Ruby XML Library
- Example Hydra implementation
- Pipelines for "getting stuff in" (digitization)
- Map Solr to MARC fields
- Sept 2011: Hydra no build-in authentication for workflows
- Fedora has both Flash & Java client based administration controls
- Fedora uses ITQL, SPARQL (others?)
- Berlin Benchmarks for triple-store performance comparisons