In a recent thread in Veeam forum, a user was complaining about slow execution of a Reverse Incremental backup saving a VM with Exchange 2007.
After a check, the user was in this situation:
– vmdk of the Exchange 2007 server is 700 Gb, with about 650 used
– every daily increment produces 200 Gb via CBT
In this situation, choosing a Reverse Incremental backup is worst even than a daily full backup: with this method Veeam have to identify in the previous VBK files those 200 Gb to be substituted and extract them, then write the new 200 Gb inside the VBK, and finally write the new VRB file. In total, there are 600 Gb moved in the backup storage, basically the same amount of a full backup made every night!
Moreover, in a server with high IO like an MS Exchange of that size, a reverse incremental requires more time than a full, and this means a bigger snapshot on VMware than an incremental backup; the side effect is the snapshot consolidation requires a high amount of time to be completed, even many hours.
So, do your calculation carefully before choosing Reverse Incremental.
I couldn’t agree more with your logic. I am even now of a mind for always using incremental with a transform every day once the backup is complete. This means less time the VM’s are in a snapshot state, quicker backups from the disk and once completed takes care of the transform without impacting any production systems. Just the snapshot commit time is worth doing it this way especially for busy servers like SQL and Exchange 07/2010.