Index structure?

I am trying to understand the structure of EV indexes. In particular, do index files grow in size as new items become indexed (and thus go through mutiple versions of each file), or do they simply multiply, with new information going into fresh files? I cannot seem to find any information on this question. I note that under each Index folder, there are a number of folders of the with names such as 11AC8AB0CDA3D404AA4E272FCBD0E2D421110000SiteName and within each folder there are a large number of smallish files. The file names within each folder have similiar names, although some folders contain more of these than others (By no means random - either 107, 142, 252 or 258 files are found in each of the 14 such folders). We only have about 40 users enabled for archiving, but in a few weeks will have 4000! So what will happen - will the files become more numerous, or will they each grow is size (or both).

This is not just a matter of ideal curiosity - I want to get some idea of what will have to be backed up. Millions of small files or thousands of larger, ever changing ones!

Any information will be appreciated.