As a user of WDQS, I want my query to be resolved in the most efficient way possible so that I can find what I'm looking for without having to worry about technology limitations. As a WDQS developer, I want appropriate queries to be offloaded from the main WDQS service so that the WDQS service doesn't break.
WMDE has done a prototype of a query manipulator that identifies a subset of queries and serves them from different services or rewrites them in order to take load off of the Wikidata Query Service.
See their demo presentation here:
This ticket is to evaluate what is required to make this prototype production-ready, including:
- Put dummy data in front of the prototype solution to see which queries go where, to see if setting up an instance with just truthy data makes sense.
- Determine how to provide a way of separating truthy queries
- Determine how to provide a way of separating linked data fragments
- Run experiment either live on production or with deprecated data from production
- Determine whether the query manipulator offers enough advantage to be worthwhile
- If worthwhile, determine whether it is feasible to run the query manipulator in a production environment
- If it is feasible, create required sub-tasks to accomplish it