Backup Copy Job to offsite Data Domain

[archibaldicus]

We are writing backups to an EMC Data Domain in our primary datacenter. The DD has been configured with a Veeam Gateway. We would like to run a backup copy job to an offsite Data Domain in order to have offsite backups. For this purpose, another Data Domain and Veeam Gateway have been setup at the offsite location. The available bandwidth between the 2 sites is 50 Mbps. For a month or so, I ran both DD’s next to each other (in the same site & VLAN) to seed some backups and to test the backup copy job.
The data copied across comes from a Windows Agent backup of a file server cluster (so no virtual machines) and is about 7 TB spread across 3 backup copy jobs. I notice that a full backup copy job seemingly moves across all data in not deduplicated form, as in Cacti 50 Mbps goes into the target Data Domain’s gateway, and 20 Mbps goes out to the target Data Domain. Furthermore, it moves the data across twice: once for each of the 2 file server cluster members. Of course, this makes the backup copy job impossible to finish within an acceptable period of time.
I have opened an SR with Veeam and they told me that WAN accelerators will not help because I have too much bandwidth available, adding that there is basically nothing much I can do. Even turning off the “Read the Entire restore point from source backup…” would not help as this would shift the workload to the Data Domains that, although unparalleled in data deduplication, have got very poor read throughput.
Does anyone have any experience with this kind of issue? Any tips would be greatly appreciated.

Kind Regards,

Stijn

[foggy]

I notice that a full backup copy job seemingly moves across all data in not deduplicated form, as in Cacti 50 Mbps goes into the target Data Domain’s gateway, and 20 Mbps goes out to the target Data Domain.

This is expected, since data travels through the gateway server, so is rehydrated. If the issue is just the initial run, you could use regular job to back the file server up to the second array as a seed and map the backup copy job to this backup then.

[archibaldicus]

Hello,

Before moving the second array offsite, it was located next to the primary array to seed the backup copies (for about a month). Are WAN accelerators no option with a Windows Agent backup copy job?

Regards,
Stijn

[foggy]

Ah, so it is already moved offsite - I thought we're talking about initial job run. You can try WAN accelerators to see whether they help. At least they will prevent duplicate data from two nodes over WAN. What are the bottleneck stats for this job?

[archibaldicus]

Hello,

The bottleneck is the throttling - I had to set it to 50 Mbps or other stuff stops working . However, the Veeam support engineer told me that I have too much bandwidth available for WAN accelerators to be effective. Furthermore, how can I calculate the size of the cache? The Veeam documentation only mentions 4 GB per OS, but this is a Windows agent backup that only contains data volumes an no OS.

Regards,

Stijn

[foggy]

For WAN cache sizing please refer to our best practices guide and the corresponding user guide section. Thanks!