More complexity... more security issues...
So now you're driving some sort of offline mode where critical data is stored locally on an inherently less secure mobile device - and has to be re-integrated back with the existing data stored on the LAN when the device returns to connectivity.
This is a big enough challenge when you just have multiple people working individually in collaboration on a single dataset on a traditional network and look at the weird workarounds that exist for that, like file locking/read only mode access alerts and such. Even with that - I've seen roaming offline profiles cause problems anyhow in those scenarios. For example, two people are working in the same document, but one person is accessing their offline copy of the document. Person A, connected, opens the file, it isn't *locked* because there isn't a connection to person B. They make modifications, save the document and go on their way. Person B accesses their cached version offline, makes modifications, save the document, and later comes into the office. In the meantime, person A has gone offsite, opened the cached document and made changes as well. Person B comes in, their offline data syncs - their updates are captured. Person A comes back in, and their updates overwrite all of person Bs updates... then the nightly backup takes place.
The next morning, person B connects, and all of their work is gone. Person A says, "everything I did, even after I left - is still there"... IT gets a call, looks at the backup, and only person A's data is on the backup...
SNAFU. I mean, the second example is a risk with any mobile device, including laptops - and happens for sure. But a lot of time real "cloud" based devices only store the local copy cached until they can get it back to the server - so it is more likely that in the example above, person B won't have a persistent copy on their device...
I think the difficulty writing and reading the above example shows why it is such a problematic thing to deliver in IT... offline documents lead to document loss - because the rules can't guarantee that it will always be the most current data that gets written when there is contention between two different examples of the same file or document.
Contributr 
