Backing Up VMs every hour

[HFX902]

Hi Community,

We are looking to backup VMs every 1-2 hours. The total size of the job would be about 800gb. What is the best veeam feature to use to accomplish this goal?

Veeam Backup and Replication 9.0
Vsphere 6.0
Esxi 6.0

[nielsengelen]

You can do this with a normal backup job. Which type of storage do you have underneath your production?

[foggy]

The major concern would be whether your setup allows to complete the number of changes occurred inside VMs every hour.

[HFX902]

You can do this with a normal backup job. Which type of storage do you have underneath your production?

Currently the job is taking over an hour at 120 MB/s

Backing up to a QNAP NAS 4tb drives x12 7200rpm 1gb NIC

[foggy]

Do you mean the full backup takes more than an hour? Incremental runs should be faster, depending on the backup method. What is the major bottleneck reported by the job?

Also, I believe Niels was asking about the production storage, where VMs reside.

[HFX902]

Bottleneck is the source

Incremental backups

[foggy]

Currently the job is taking over an hour at 120 MB/s

Still not clear, whether it is for the full or incremental run.

What transport mode is being utilized? Direct SAN, hotadd, network?

[HFX902]

Incremental, Transport is network

[foggy]

You should consider using either hotadd or direct SAN, if applicable. Network is the slowest one.

[HFX902]

You should consider using either hotadd or direct SAN, if applicable. Network is the slowest one.

Thank you for your feedback. Along with backing up every hour we would like to have our restore time fast as possible, would reverse incremental be that best option for fast restores?

[vClintWyckoff]

Yes Reverse Incremental would be the fastest, however the 7200RPM QNAP NAS might not be able to handle the Reverse Incremental IO requirements. During RI backup mode we'll inject the changes into the latest VBK and then merge the replaced blocks into the VRB creating 3x the IO compared to a Forward Incremental.

[foggy]

Though difference in performance can be observed only on a significantly long chain.

[slos]

Foggy, can you comment on the restore point length of ‘significant’? 7 -21-60 points ahead of the last full?

HFX, for restore performance the Hot Add proxy is the most efficient choice. For backup performance you’ve noted that your using a QNAP NAS, how are you exposing the discs on that piece of hardware for Veeam write data to it? For example, do you have the device added to Veeam as a Shared Folder?

[foggy]

Just to clarify, I meant performance difference between forward and reversed incremental modes. Extremely long chains can potentially impact restores from long forward incremental chains due to metadata handling, however, it depends on the storage capabilities as well, so there's no general rule of thumb. Chain length can be limited due to safety reasons as well.