"CN" <cnliou9(at)fastmail(dot)fm> writes:
> I thought I have got no more question in this thread, and I was wrong :-(
> Is the extra 300ms in UNION essential?
Dividing, it looks like the Append node is taking about 3 microseconds
per tuple, which seems kind of a lot considering it isn't really doing
anything. I suspect though that most of this overhead is due to the
EXPLAIN ANALYZE instrumentation, and would in fact go away when you
are just running the query.
regards, tom lane