Craig Ringer <craig(at)postnewspapers(dot)com(dot)au> writes:
> I'm quite curious about why setting the transaction to readonly helped
> its performance. Could it be to do with setting hint bits or something
> along those lines, perhaps? Anyone?
AFAIK that's actually a pessimization. Declaring the transaction READ
ONLY does not activate any optimizations that wouldn't be there
otherwise, and it causes a few more cycles to be expended to check that
each statement is allowed under READ ONLY rules. So I think that this
report is mistaken, and the performance change came from something else.
regards, tom lane