2016-07-08 Graph DB/LDP/Marmotta Review
Date
Attendees
Goals
- Identify priorities and concerns for evaluating datastore situation
- Outline work plan.
Notes
- Plan to make decision before mid-September
- Any data migration would need to take place in the late fall or early-winter.
- Issue: Staff time; trade-off between making this migration. Would we need a DBA to continue to use Marmotta w/ PSQL?
- Issue: Sysadmin concerns: Ease of installation, maintenance, scaling.
- Scale out is a high priority (ES as model here)
- Less configuration
- Issue: Is a Graph DB better suited?
- Benchmarking needs to be in scope.
- Marmotta:
- Versioning in Marmotta is not LDP-aligned.
- CG work and Fedora work http://fedora.info/spec/2016/05/20/resource-versioning#common-http-response-headers
- Lean toward preferring timely implementation of standards development, rather than retaining existing unused features.
- Other LDP additions?
- Versioning in Marmotta is not LDP-aligned.
- Issue: KriKri LDP interactions are slightly Marmotta flavored
- Refactor of Krikri::LDP::Resource and Krikri::OriginalRecord would be necessary to change server layer.
- Tom Johnson (Unlicensed) to share with Mark Matienzo (Unlicensed) & schedule followup
Action items
- Define requirements - Due July 25.
- Environmental scan
- List datastore options: Marmotta backends, other LDP services
- Initial evaluation - Pros/Cons
- Select options for further evaluation
- Develop benchmarks
- Scale Requirements?
- Initial benchmarking on current configuration
- Full evaluation of selected datastores
- Report & recommendation - Due Sept. 9.