This is a nice solution. However when we have hundred other files for similar process it will eventually lock a considerable memory forever.

I was thinking of having split the task into two

1. File watcher task (WMI)

2. File processing task (Package)

The watcher task could be a package or windows service which will be running continuously. It would trigger the processing package when the file is in place.

This way the system resources would be less consumed.

The File watcher task would be singular for all the source files. I.E. it will keep on looping through all the source directories and fire appropriate package as and when the file arrives.

Obviously it would need to have a mapping between file and package stored either in sql or config file.