Personally i hate collections as andy said, some of the reasons I don't recommend it are
1. If a collection goes bad, rather than losing one item, you can lose several dozen or even a hundred archived emails with one lost or bad CAB files
2. Processes such as EVSVR scans, index rebuilds, Move Archives, Meta Data Store builds are 25% slower, because it has to extract the item from CAB to disk etc
3. It's a one way process, theres no way to get out of collections except for cheap hacks
4. There have been multiple programatic issues with collections, such as having incorrect counts of items stored in the CAB, which is really important
5. Deleting items or storage expiry does not lead to immediate space savings, EV has to run through a collections process to check each CAB to extract the legitimate items, delete the CAB etc, so you could have deleted many archives thinking you'd save a lot of space, but not see anything happen on disks
6. You could also end up in a scenario where you need to do a restore of your Vault Store database, and the restore is from before a sparse collections was run, so it may say a particular item is in collection12345.cab, but in fact, it was placed in collection67890.cab and thus the item can't be retrieved
That along with other things that i've seen such as items being recorded in the database in wrong collections, ev losing track of the refcounts etc, its just a nasty process that isn't fun, and I don't generally recommend one way type of processes
As for your backup issue, if you're on a NetApp, why not use Snapshots?
https://www.linkedin.com/in/alex-allen-turl-07370146
