...
- KriKri Datastore Requirements
- Based on assumptions that we need LDP and SPARQL; need to support generalized schema for RDF (i.e., not just DPLA MAP).
- Scaling up vs. Scaling out - need to scale up first. We scale out if we have use cases to support it.
- If scale out is required as part of our needs, that changes some of our needs related to concurrency, etc
- Open source - part of assumption that we prefer FLOSS is based on our mission.
- Cost/benefit - internal business need trumps tech team principle if there's no other option
- Keeping smaller scale is OK as well.
- New requirements:
- MUST be a production-ready FLOSS persistence implementation that the software supports
- SHOULD be a FLOSS implementation for DPLA that provides commercial support, etc.
- PATCH would be nice
- Bulk export - subject for discussion
- High availability - not a requirement but a plus
- Scalability and relationship to throughput
- General Purpose RDF in LDP-RS - relates to behavior of LDP implementations like Fedora 4; strongly preferable
- Does it make sense for us to maintain data about resources not under our control? Don't really use them at this point.
- Reasoning - YAGNI
- Versioning - probably requires a larger discussion
- Next steps
- Environmental scan - due august 9
- Benchmarks
Action items
- Tom Johnson (Unlicensed) to follow up how to coordinate work on environmental scan and benchmarks