What size is the largest vm you backup ?

[davecla]

@Shestakov Backups use Network Mode (NBD). Backup proxy role is a physical machine and the uses local storage for the repository.

@Dellock6 - Good point on many smaller VMDK's. I'd always shied away from spanning a bunch of VMDK's to make a bigger volume, but looking from a backup perspective, It's got to be worth a test to see how the benefits shape up.

[Shestakov]

When using Network transport mode usually Bottleneck is Source.
For bigger infrastructures like yours it`s recommended to switch to Hot-Add or Direct Storage Access transport mode.
Thanks!

[stewsie]

HI

I am backing up a few VMs where the total size is around 9TB. The backup repositories are limited to a max size of 16TB so space can be a challenge. Although the backup speed is OK, the merge operation takes a long time and sometimes fails. I have been experimenting with some smaller backup jobs where I have configured them to take an active full weekly. Doing this avoids the merge operation.

Why are Active Full backups not recommended and does anyone have any advice regarding improving the merge speed for backup and also copy jobs. The storage in use for the repositories is EMC VNX 5300.

We are still running Veeam B&R 8

Thanks

[skrause]

Active Fulls are typically not recommended due to the impact on production operations. Having a snapshot open longer, and having to copy the entire content of the disk from production storage instead of just changed blocks can cause performance issues in many scenarios. But if your production systems can handle the larger snapshot and the extra throughput to do an active full there is no reason why you can't use it if it fits your environment best.

As far as increasing merge speed, many times the only way to ensure an increase in speed is to increase your spindle speed on your backup storage and use RAID10 instead of RAID5/6. Merges are the most random I/O intensive task you will have with Veeam.

[stewsie]

Thanks for the response which confirmed my thoughts.

[Shestakov]

Although the backup speed is OK, the merge operation takes a long time and sometimes fails.

Paul, what kind of merge did you have? Note that Reverse incremental, Synthetic Fulls and Forever-Forward modes require different number of IO operations so show different performance.
Thanks!

[stewsie]

Using Incremental (recommended) as the backup type. Some of the jobs are getting very large now where the full backup file is around 9TB. I am looking at using synthetic full now in an attempt to reduce the risk of the full being corrupted due to the large size.

The copy jobs merge operations can take a long time for the larger jobs

[Shestakov]

Synthetic full doesn`t help to avoid backup corruption. In that case corrupted data blocks will be part of newly created .vbk
For recoverability it`s recommended to use Surebackup or periodic active fulls.
Thanks!