Recently I noticed an “interesting” thing that I would like to share with the community. When a DPM or Azure Backup Server archive data to Azure its known that the initial online job will take some time to finish since it will ship all the data to Azure from the on-premises server.
All the data that is replicated to Azure after the initial online job will run in an optimized state. This means that only the changed data will be shipped to Azure from your DPM or Azure Backup Server and it will be finished quite fast depending on the actual data churn.
If that online job fails the MARS Agent will lose track of what is shipped to your Azure Recovery Services Vault and needs to verify the data and the history that resides within the Azure Recovery Services Vault to be sure what data has changed from the last archived Recovery Point provided to Azure.
This is a very time-consuming job and it can take over 100 hours to complete, just had a client that had an online job running for forever and looking at the job history in the DPM console the last Online job had failed. This means that the DPM or Azure Backup Server needs to perform an Unoptimized online job.
I hope this article sorts out some question marks for you 😊