In other scenarios we've found it necessary to split this query into two pieces:
- first query for all observation URIs and spill the results to disk
- then page through the URIs and insert then into a second query using a
VALUES clause
This should scale better than e.g. LIMIT/OFFSET (which still requires the database to assemble the offset solutions).
In other scenarios we've found it necessary to split this query into two pieces:
VALUESclauseThis should scale better than e.g.
LIMIT/OFFSET(which still requires the database to assemble the offset solutions).