as a general rule, you should try to find-out if your script is CPU bound (so there is some algorithms to be checked) or IO bound (waiting for DB to finish or waiting for the HDD). Just check the CPU load or use NYTProf.

Also it seems to me a very scalable task, IOW you can run the script which find all the files to be managed and split this list in lists; then fork as many times for every list and process the data. This will solve latency issues to the db and speed-up the other cases also.