Backing up of the fileshare to Azure Recovery Services Vault (https://azure.microsoft.com/en-us/services/backup/)
1. Initial backup was 1.5TB, mostly in small files. I won't even hazard an estimate of how many files there are. This backup took 2 weeks to run, we only throttle during working hours. I will not want to do this again.
2. As expected, once the initial backup completed, the incremental backups fell within 20GB. I conclude that the majority of our data is static.
3. During the initial backup the counter would move 50GB per day, yet during the incremental backups a 20GB transfer would take 60 hrs. What is going on? The performance meter shows a sizable amount of disk activity, but no bandwidth used. This I attribute to the nature of "lots of small files". There is a lot of churn when computing hashes for many files, a lot more than a similarly sized few large files. As CPU, disk, RAM and network bandwidth are not excessive, this performance issue may be difficult to solve.
4. Got a shock one day when the vault usage dropped from 1.5TB to 40GB. The counter stayed there a while, but has returned to it's former value. How should I verify that the data I took so long to back up is still there? Am contemplating setting up ABS in a temporary Cloud VM, and restoring the entire contents there.