Re: Thousands of partitions performance questions

2019-04-29 Thread David Rowley
On Mon, 29 Apr 2019 at 19:20, Shai Cantor wrote: > Some notes I haven't shared or were not clear enough on the previous post. > > Data is inserted using the copy command only and in an offline manner. > Meaning, no user action creates or updates the data. An offline job runs is. > Number of inse

Re: Thousands of partitions performance questions

2019-04-29 Thread Shai Cantor
Thanks a lot for your feedback. We're gonna user PG 11 on AWS RDS. I will do some workload simulations. Some notes I haven't shared or were not clear enough on the previous post. 1. *Data is inserted using the copy command only* and in an offline manner. Meaning, no user action creates or up

Re: Thousands of partitions performance questions

2019-04-28 Thread David Rowley
On Mon, 29 Apr 2019 at 17:13, Shai Cantor wrote: > Will the db hold 135000 (45000 * 3 months) partitions under the assumption > that I query only 1 partition? > Should I model it differently in terms of schema, partitions etc.? Which PG version? Before PG11 the query planner pruned unneeded par

Thousands of partitions performance questions

2019-04-28 Thread Shai Cantor
- *General* - Our system gathers data from CI builds regarding a customer's code base - *Data includes* - type - methods/lines/branches A.K.A - code elements - files - *Queries* - The data is queried for a specific build only - Queries are