Comment by supermatt

very simple processing, not memory bound, tiny data-set - of course its going to be faster locally when the command itself takes less time than the networking, distribution, coordination and collation overhead of using any distributed tool...