Azure VM incremental backup identification of changes
We have configured daily Azure VM backup in our landscape. The DB VMs have one attached disk which stores the DB backup temporarily before it is moved to a cheaper blob storage location. This DB backup and transfer happens each day and is completed and space cleaned up before the Azure Backup kicks in for that VM. So ideally when the Azure backup runs each night by then the disk is almost empty except for some log files which get created by the DB software which we need to backup by Azure VM backup. We are finding the the instant snapshot for this particular disk during Azure backup is taking up a lot of space. It is a 1 TB disk but as I said mostly it gets emptied before Azure backup starts. Why is the snapshot so large when the disk has been emptied? I know Azure incremental identifies changes using block level checksum calculations. Does the checksum differentiate between two different sets of empty blocks – I mean the empty block that was created by deletion of last backup and the new empty block created during the current DB backup deletion? Any insights much appreciated.