Open didip opened 1 year ago
Hello, I also noticed a very high bump in insert-time-per-record when bumping amount
to 1B from 10M (as in, more than 100x time for 100x records). While the mass-loading pattern doesn't exactly complement a store built for concurrent DML, I agree the 1B benchmark should insert in a reasonable time regardless - especially with all data being known before hand.
Indeed, insertion is a bit slow as it touches most columns. I have a few questions to clarify this.
I modified the million benchmark to billion and it took forever to insert.
Is it possible to expand the commit.Buffer to speed up the transaction?