Greg Stark <gsst...@mit.edu> writes: > The backwards scan is awful for rotating media. The reading from the > end and writing to the beginning is bad too, though hopefully the > cache can help that.
Yeah. And all that pales in comparison to what happens in the indexes. You have to insert index entries (retail) for each moved-in tuple, then after doing the intermediate commit you run around and remove the index entries for the moved-off tuples. Lots of nonsequential access to insert the entries. The cleanup isn't so bad --- it's comparable to what regular lazy VACUUM has to do --- but that's just one step in a very expensive process. regards, tom lane -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers