When creating the original approach to iterate through all articles in a wiki to find new suggestions with a maintenance script, @Urbanecm_WMF made the good point that our current database queries might be somewhat inefficient by checking for every page individually whether they have suggestions in our GrowthExperiments database table growthexperiments_link_recommendations.

As we are now rolling out the new way of refreshing those Link Recommendations to more wikis, we should investigate if batching these queries for ~500 pages at a time would be better than doing them individually. This is about the database query coming from LinkRecommendationStore::getRecommendationStateByRevision which goes to the primary database.

Acceptance criteria:

• Investigate if there is benefit from batching these queries for by 500 at a time when checking for whether they are known in the growthexperiments_link_recommendations table
• Investigate whether with the new way of doing things (iterating through all the articles in a wiki), we actually still have to query the primary database
• if we can improve things here, outline a plan of how to implement those improvements and estimate the effort required