Yes that was the point of my second question, for each record in each xml file are you handling querying the database, determining what action to take and then sending...

Do you have an easy way to identify what format each xml file, like file naming convention or does each format go to separate directory or the like?

One...

Sure you might be able to reduce the total size of your table and yes that would theoretically decrease the the maximum size of your database and backups and if...

Why do you even need the cross join?  Or are you asking how to do that without using a cartesian product, if so what is the problem with that? 

You can try an opensource document management system that'll run in a LAMP setup and a cheap cloud linux server to run it on.  I personally can't recommend any but...

It really depends on what the application is doing.  At the basic level all execute process is doing is starting an OS process that runs whatever command you gave it...