In addition to that, a following case would be solved with this approach: When many processes are referencing many tables defined foreign key constraints thoroughly, a huge amount of memory will be consumed regardless of whether referenced tables are partitioned or not.
Attached the patch. Any thoughts?
Amit Langote has done some great work at eliminating SPI from INSERT/UPDATE triggers entirely, thus reducing the number of cached plans considerably.
I think he was hoping to have a patch formalized this week, if time allowed.
It doesn't have DELETE triggers in it, so this patch might still have good value for deletes on a commonly used enumeration table.
However, our efforts might be better focused on eliminating SPI from delete triggers as well, an admittedly harder task.
Amit's patch is now available in this thread [1]. I'm curious if it has any effect on your memory pressure issue.
