Re: Really bad insert performance: what did I do wrong?

Kevin White <kwhite@digital-ics.com> writes:
> My box, however, isn't a production server, so I attempted to create the
> database again on a Sun Blade:
> ...
> It took almost 2 days to load the table on this box.

Yipes.  We found awhile ago that Solaris' standard qsort() really sucks,
but 7.3 should work around that --- and I don't think qsort would be
invoked during data load anyway.

Do you want to rebuild Postgres with profiling enabled, and get a gprof
trace so we can see where the time is going?  You don't need to run it
for two days --- a few minutes' worth of runtime should be plenty.

            regards, tom lane