medium10 pts · first answer countsquery-tuningpostgresql

Counting a huge table quickly

A dashboard runs SELECT COUNT(*) FROM events on a 2 billion row PostgreSQL table every minute and each run takes minutes. An approximate number is fine. What is the best option?

A.Add a B-tree index on every columnB.Read the planner estimate, for example reltuples from pg_class (kept fresh by ANALYZE and autovacuum)C.Run VACUUM FULL before each countD.Use COUNT(1) instead, which skips reading rows

← Terraform plan wants to replace the DB Take the medium mock interview →