On Thu, 2024-02-01 at 10:23 +0100, Dirk Krautschick wrote: > I have run a test with pgbench against two cloud vendors (settings, > parameters almost the same). > Both Postgres (or whatever they do internally when they call it as Postgres > offering, NOT Aurora or so :-) ) > > I have got a strange result that cloud vendor 1 is performing almost > everywhere better in matter of > read and write but except in the init phase of pgbench it took almost double > the time.
Nobody except those vendors could tell you for certain, but perhaps on the one system the initial data load is fast, because you have not yet exceeded your I/O quota, and then I/O is throttled. Yours, Laurenz Albe