My dataset is 100K+ records with geo field and userId this records belongs to.
I’m designing two queries (one for all records, another by userId) that would return either full documents bounded by a GEO box, or clusters with average lat/lng of a subset with count. Wanted an opinion on what approach to take for better performance.
Should I create a MapReduce view and pre-calculate clusters in there? Should I leverage N1QL with geo index grouping by rounded lat/lng to 2 decimal points?