Re: BUG #13484: Performance problem with logical decoding

On 2015-07-06 19:29:30 +0200, olivier(dot)gosseaume(at)free(dot)fr wrote:
> To be more concise :
> - ONE transaction with 4095 operations -> consumed in 80mS
> - TEN transactions with 4095 operations each (so 40950 operations) -> 380mS (which as you said is very good -less than linear growth-)
> - ONE transaction with 4096 operations -> consumed in 4204mS (ouch ...). I confirm there are only 4096 changes in one transaction
> - TEN transactions with 4095 operations each (so 40950 operations) -> 34998mS, ouch again

Given the third result it's unlikely, but just to make sure: Are you
sure network issues aren't a factor here? 40950 changes are about 3.8MB,
over a slow network that can take a second. Additionally psql will
display all the changes in a pager, which'll not be fast either (but
shouldn't be included in \timing's output).

Greetings,

Andres Freund