You SHOULD by default only use one, just to simplify your recovery workflows. What benefits are you achieving by using both solutions? Just to identify what is making you currently use both.

Should there be a benefit that makes you want to use both, then you should do something such as one of them only performing “crash consistent” backups of the server.

Regardless of choosing copy only options for transaction log processing in Veeam. You still can’t use them reliably. For example:

If MABS uses VSS integration and truncates snapshots, and Veeam doesn’t truncate logs/copy only:

If Veeam runs first for a daily backup, the moment that MABS performs a backup, it will mark all transaction logs as free for reuse (log truncation), and so the newer transaction logs should be used from the backup that MABS created to roll up the transactions.

Inversely if MABS runs first and then Veeam performs a copy only daily backup, your transaction logs would be good until the next time that MABS performs a backup + backup log truncation.

Thats as far as my knowledge is aware. I don’t doubt there are ways to work with transaction logs and database states from different points relative to the backups taken. But I’d be wanting to do that with Microsoft support exclusively as you’re playing with two different database types here (JetDB and SQL), what works for one might not be anywhere near similar to the other.

Hope this helps, and because it’s a complex topic, I welcome anyone to correct anything I’ve potentially got wrong here!

I use dpm for backup sync every15m and veeam for a full VM Backuo one a day. What is the best practice for backup Exchange Dag whit only veeam? And the same things and question hav i for sql server

Thanks a lot

Alfred 

Yes as Michael mentioned one solution should be used and if you are using DPM for 15 min sync then maybe you look at CDP in Veeam as it can do a similar thing and it is all one solution.

For Exchange DAG it’s recommended to backup the passive copies of the databases to avoid failovers, handy if you’ve got an active/passive setup. Otherwise you CAN use either VM level backups, or agent level backups. Generally agent is preferred to avoid VM stun times, this can be mitigated with varying success based on DAG stress, datastore performance, and hypervisor + version, if you’re using vSphere for example, then newer ones are much better at snapshot stun times.

SQL is more nuanced depending on standalone, Always ON or failover cluster.

Veeam and MABS can do the same thing of backing up locally and then to Azure. MABS will use recovery vault, Veeam v12 can use immutable object storage.

A question if you’ll let me: where do you intend to restore these workloads to if you need to? Purely in case you didn’t know, you can’t send emails directly from an Azure VM, you have to use a mail server to relay your email.

SQL a have a cluster with 2 VM with shared disk, i backup the VM with veeam and what about the transtionlog…...or better what the best practice to do this?

As in a CSV? If so Veeam doesn’t process CSVs (or at least the agent does, would have to check at VM level). If you are using CSV I’d try the Veeam SQL Server Plug-in to process the data, and then use VM or Agent for the base-OS pieces

yes is in a CSV, i don’t can save only the VM?

https://www.veeam.com/wp-sql-alwayson-availability-groups-virtual-environment.html
 

have a read of this and see what applies best to your environment

The agent processes the data on the CSV.

But you cannot backup the CSV itself and cannot create a snapshot of a VM with a CSV.

So, you have create an empty VM  and create and attach the CSV. Then boot it with the recovery media to restore the OS and your data.

Thanks for elaborating, I was grabbing the CSV limitation from here: https://helpcenter.veeam.com/docs/backup/agents/cluster_support.html?ver=120

• Backup of CSV (Cluster Shared Volumes) is not supported. Cluster disks used as CSV are automatically excluded from backup.