Re: select distinct runs slow on pg 10.6

[Rick Otten <rottenwindfish@xxxxxxxxx>]

On Wed, Sep 11, 2019 at 12:38 PM Dinesh Somani <dinesh@xxxxxxxxxxx> wrote:

I think Merlin has outlined pretty much all the options and very neatly. (As an asides Merlin could you possibly elaborate on the "C Hack" how that might be accomplished.)

To OP, I am curious if the performance changes were the query rewritten such that all timestamp columns were listed first in the selection. I understand it might not be feasible to make this change in your real application without breaking the contract.

Regards

Dinesh

It looks like AWS has a pgbouncer query re-writer service that might be a starting point:

https://aws.amazon.com/blogs/big-data/query-routing-and-rewrite-introducing-pgbouncer-rr-for-amazon-redshift-and-postgresql/

I've never used it.