On 2016-08-22 13:54:43 -0400, Robert Haas wrote:
> On Sat, Aug 20, 2016 at 11:17 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> > I'm inclined to suggest you forget this approach and propose a single
> > counter for "SQL commands executed", which avoids all of the above
> > definitional problems. People who need more detail than that are
> > probably best advised to look to contrib/pg_stat_statements, anyway.
>
> I disagree. I think SQL commands executed, lumping absolutely
> everything together, really isn't much use.
I'm inclined to agree. I think that's a quite useful stat when looking
at an installation one previously didn't have a lot of interaction with.
> Haribabu's categorization
> scheme seems to need some work, but the idea of categorizing
> statements by type and counting executions per type seems very
> reasonable.
I'd consider instead using something like COALESCE(commandType,
nodeTag(Query->utilityStmt)) as the categories. Not sure if I'd even
pivot that.
Andres