Postgres 12, 13, 14 and 15 vs the Insert Benchmark - not many performance regressions, Postgres remains boring (for me).
smalldatum.blogspot.com/2022/06/insert
Conversation
Replying to
I appreciate your work on this - regressions like this matter. The regression in 15 will be investigated soon.
1
8
Replying to
Thanks! Not related to the regression, but: it would be nice to see final index sizes for insert benchmark (maybe this can be available the next time around?). A big focus of mine these days is efficiency. (Partly due to being influenced by your work, actually.)
3
1
There are competing factors that either hurt or help latency as we do more to control index bloat. Not necessarily because of added cycles per se. Some regressions are a natural consequence of smaller indexes; the concomitant increase in latch contention is the problem.