VIVO-983: Test VIVO against multiple triple stores

[chenejac]

Ted Lawless (Migrated from VIVO-983) said:

To produce some basic performance numbers that the community can use to make decisions about backend triple storage, execute and record time for completion of the following tasks against multiple SPARQL 1.1 triplestores.

• ingest sample data
• run full inference

A small people, positions, and organizations dataset is available here: https://github.com/lawlesst/vivo-sample-data.

Possible triple stores to test

• Jena SDB with MySQL (default)
• Fuseki and TDB
• Sesame
• Stardog developer edition
• AllegroGraph
• Apache Marmotta Kiwi
  • BlazeGraph

[chenejac]

@brianjlowe: Is there a performance test between Jena SDB with MySQL and TDB? How did the group make decision to move VIVO 1.12. to use TDB as a default option?

[brianjlowe]

@chenejac See https://wiki.lyrasis.org/display/VIVO/Performance+Testing

The default triple store setting in applicationSetup.n3 was actually switched in VIVO 1.11. If I recall correctly, the decision was based on a number of factors: (a) lack of ongoing support for SDB, (b) performance figures, (c) ease of installation (no need for an RDBMS) and (d) maturity of TDB compared to TDB2 based on a meeting of the team with Jena's Andy Seaborne.

[chenejac]

@brianjlowe Thanks for URL and information. Although not all options have been tested, I believe the wiki page might be used by implementers/adopters to decide which option they will use (or to keep default option). Therefore, I believe this issue might be closed.

[awoods]

Minimal notes from the meeting with Andy Seaborne mentioned by @brianjlowe : https://wiki.lyrasis.org/display/VIVO/2019-12-06+-+Special+Topic+-+TDB+vs+SDB