On Thu, Jan 17, 2019 at 9:19 AM Ben Snaidero <bensnaidero@geotab.com> wrote: > Any ideas as to why this is happening?
Not really, I would expect roughly double execution time, not an exponential increase. Still not experienced enough to diagnose with what has been provided but I will suggest you provide the version that you are running these queries against and confirm that the results are consistent across multiple runs to discount the possibility that locking or other transient work is involved.
Curious if increasing work_mem helps but its mostly a trial-and-error thing for me (I wouldn't expect it to)...
David J.
Query was tested multiple times with similar results. I also tried increasing work_mem with the same results as well. Memory did not increase or even come close to the 10MB setting. Query does seem to occupy one full cpu for the duration of the run though. Query is running on windows with Postgres 9.6.7