I haven't read the whole exchange (sorry!), but if your reason for doing this is to weed out duplicates then I suggest you do that directly in SQL.

Something like

select distinct key, value into unique_table
  from duplicate_table
[download]

should work, and shouldn't tax your "monster dec alpha box" excessively.

If you want to find which rows have duplicate keys you may have to add a COUNT(*) and a GROUP BY clause...

Michael

Update Note that the select statement above will only work if you have "create table" priviledges in the database...