Symptoms

You have 2 or more backup plans applied to a Hyper-V virtual machine. You notice that backups start taking much time.

When you check the service process logs for the affected plan, you see the following records, where "bytes read" values are too high and nearly match the total size of the virtual disk:

2019-10-03T23:39:48:564+07:00 6304 I00000000: I/O statistics for device 'VM(IFACE_ATA0:0)': 109139525632 bytes read, 0 bytes written.
2019-10-03T23:39:48:658+07:00 6304 I00000000: I/O statistics for device 'VM(IFACE_ATA1:1)': 1820552261120 bytes read, 0 bytes written.

Cause

CBT calculates differences since the most recent backup. If multiple backup plans are applied to the machine, it is likely that the most recent backup was initiated by another backup plan. CBT sum calculation fails and all data is saved into the archive instead of changed blocks only.

Solution

You can specify multiple backup destinations in one backup plan, this feature is called backup replication. See details here.

If CBT calculation works correctly, you can see the following entries in logs:
2019-10-04T16:16:12:017+07:00 18352 I00720000: Changed Block Tracking (CBT) information has been successfully retrieved: 4761781248 bytes are changed. Change ID: 'BD0024AF-C362-4170-BE4C-39B7CC93643E'