hard20 pts · first answer countspartitioningpostgresql

Deleting old data every month

A logs table gets 300 million rows a month and must keep 12 months. Monthly DELETE jobs cause huge WAL, bloat and replica lag. What is the best design?

A.Run the DELETE in one big transaction at nightB.Range partition by month and DETACH (optionally CONCURRENTLY) and DROP old partitionsC.Move to VACUUM FULL after every deleteD.Disable WAL for the logs table using a session setting

← Unique email on a partitioned table Primary demoted when etcd is unreachable →