1) Replace the max date check with the file's update date, resulting in one fewer read.
2) Use basic Exists/Not Exists statements (with the field name being identical) to allow for optimized QVD loads. Using two different field names breaks the optimization.
3) Optimize your SQL queries if they aren't already optimized. It looks like you're loading the entire history, where you could be loading only things that have been updated if you have a last-update type field in your table.
