On 10/1/2022 13:56, Julien Rouhaud wrote:
> On Mon, Jan 10, 2022 at 12:37:34PM +0500, Andrey V. Lepikhov wrote:
>> I think, pg_stat_statements can live with an queryId generator of the
>> sr_plan extension. But It replaces all constants with $XXX parameter at the
>> query string. In our extension user defines which plan is optimal and which
>> constants can be used as parameters in the plan.
>
> I don't know the details of that extension, but I think that it should work as
> long as you have the constants information in the jumble state, whatever the
> resulting normalized query string is right?
Yes. the same input query string doesn't prove that frozen query plan
can be used, because rewrite rules could be changed. So we use only a
query tree. Here we must have custom jumbling implementation.
queryId in this extension defines two aspects:
1. How many input queries will be compared with a query tree template of
the frozen statement.
2. As a result, performance overheads on unsuccessful comparings.
>
>> One drawback I see here - creating or dropping of my extension changes
>> behavior of pg_stat_statements that leads to distortion of the DB load
>> profile. Also, we haven't guarantees, that another extension will work
>> correctly (or in optimal way) with such queryId.
>
> But then, if generating 2 queryid is a better for performance, does the
> extension really need an additional jumble state and/or normalized query
> string?
Additional Jumble state isn't necessary, but it would be better for
performance to collect pointers to all constant nodes during a process
of hash generation.
--
regards,
Andrey Lepikhov
Postgres Professional