My sys ops team have recently handed over a brand new set of postgres clusters (prod, stage, test etc). We don't have any DBAs and they are under resourced. Mistakes in config have been found before. I'm not a DBA but I understand database concepts and I am happy to do some learning in my own time.

My principle concern is whether I am getting the expected level of performance given the hardware deployed (its all on prem).

I know that different applications will see different performance based on their queries and the stats of the dataset etc, but at least I would like to verify that the hardware, OS and postgres have been correctly installed and configured.

Where should I start? Ideally I could run some perf tests and at least verify that all the instances perform similarly. Is there a set of standard performance tests that I could run?

  • "Ideally I could run some perf tests" - postgresql.org/docs/current/pgbench.html – a_horse_with_no_name Jul 29 '19 at 7:43
  • Ok, I am able to run some tests using pgdump. But which test parameters is somewhat arbitrary and how to interpret the results? Is there a set of existing test results that I can compare against? – ThatDataGuy Jul 31 '19 at 9:42
  • You interpret the results to see if your different environments deliver the same performance. As those are your servers, obviously there is nothing you can compare that against. – a_horse_with_no_name Jul 31 '19 at 9:44
  • Not even a rough order of magnitude? Eg, pgbench -c 10 -T 60, gives 10 / 100 / 1000 / 10000 TPS ? – ThatDataGuy Jul 31 '19 at 11:55
  • One computer gives me tps = 1721.781056 one gives me tps = 656.818495 one gives me tps = 492.373199 and one gives me tps = 6942.873258 - what does the tell you now? – a_horse_with_no_name Jul 31 '19 at 12:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.