Database Normalization - File Inventory - Change tracking

Let's say I have 1000 computers, all nearly identical builds, users add and change things here an there, but overall very identical boxes. Let's assume I'm collecting the path, name and file version for each file on a users PC. How do I create an efficient index, I don't want to store c:\windows\system32\notepad.exe a thousand times in the db, I would think each machine could have a pointer instead to that files entry (granted the path/version/name match, lets assume they do)
The concept is as follows...
machine01 has files: 1-200, 344, 421-500, 512-566, 578, 590, 600-700 etc...
Or should it go the other way
notepad.exe is on machines: 1-88, 90-95, 100
calc.exe is on machines: 2-41, 42, 44, 48-90, 92-100

Each machine writes a local inventory, ftp's the file to a server that add's the files/path/version to the db, and would then be able to through and update pointers after all the file data is entered.
Well I guess I'm kind of answering my own question, if I want to search for machines with calc.exe on them, the second example is clearly better... I'm going to ask the question anyway in case I have no idea what I'm talking about.
I'm also wanting to track the changes... machine01 had files 44, 67, 99 changed to 1002, 1003, 1004, and files 2, 19 and 55 deleted. I know I'm all over the place with this question, but it should convey the idea.
-rich