the row estimate became ~1000x smaller with the stat in place, so it
looks like it grossly miscalculates the query plans without the stats
for large tables representing M:M relations.
On Wed, Feb 2, 2022 at 11:47 AM Michael Lewis wrote:
>
> What does the row estimate look like on the scan of data
I recently moved a postgres DB to a more powerful server with 1TB of
RAM instead of 64GB before. To my surprise after running the tuning on
the new server, the performance of joins deteriorated by 4000x
compared to the old server. I carefully examined all of the changes
and found the culprit:
if I
