Jeff Boes <jboes(at)nexcerpt(dot)com> writes:
> Given a fairly large (7 MRows or so) table with more than one index
> (say, two): would it be theoretically faster to do
> REINDEX TABLE "foo";
> or
> REINDEX INDEX "ix_foo_one";
> REINDEX INDEX "ix_foo_two";
I doubt you could possibly detect any difference. The indexes are
rebuilt one at a time in any case, and the statement-level overhead
is insignificant.
regards, tom lane