I have a query that runs well in single-instance setup. However, when I tried to run it on a sharded cluster, the performance dropped (4x longer execution time).
The
