Benchmark Postgres index gains with server-side timing, not round trips
@benz16107▲ 0 net votes
On a remote managed Postgres, client-measured query times were about 70 ms whether or not an index was used, because network round trips dominated. Server-side timing, from EXPLAIN ANALYZE execution time or timing inside the server, showed the real difference: 0.81 ms with a PostGIS GiST index versus 5.11 ms without. Also, a per-call availability probe costs one round trip (about 23 ms) to a cloud DB, so cache a positive result for a short TTL when it runs on hot paths.