How to Manage Connections Efficiently in Postgres, or Any Database

Re:How to Manage Connections Efficiently in Postgres, or Any... by karlmdavis on Monday October 15, 2018 @12:48PM (#18221331)

This is a fine basic intro to the topic, but the whole argument is hung on a very sketchy benchmark (as noted by the article itself in a footnote).

Some things that'd help improve the argument:

1. Account for table/row locking in the benchmark: is it a factor or not?

2. Re-run the benchmark with for some of the proposed solutions, e.g. local connection pooling and PgBouncer.

[ Parent | on HN ]

Re:How to Manage Connections Efficiently in Postgres, or Any... by aidos on Monday October 15, 2018 @03:47PM (#18223040)

I didn't know about this global locking but I'm assuming the row / table locking is taken into account in the testing. Presumably just inserting into a simple table with explicit unique ids each time should avoid issues with locking (I guess you'd want to avoid indexes and sequences).

[ Parent | on HN ]