It depends on where you are getting this data. Are you just creating an XML file just for the sake of creating an XML file? Is there an actual reason that you need it to be XML? If not, you are probably better off with a database, anyway. Also, you should be using cache to your advantage all the time.
As for your initial question, I am not exactly sure what it would do in that case; you should run a quick test to find out.
Thanks for the response! My situation is that nearly 90% of the time a script is "Selecting" data from the database to display for the user. During this data's lifetime it may be selected close to 200 times a day, times this by numerous amounts of other data sets and you can see my DB being taxed.
I figured a good way to proceed might be to store this information in XML, have the script parse and use data from the XML. When a change occurs with the data a python script could re-write the XML.
The large XML files occur when a summary of all the data sets are compiled.