Backup to AWS/Azure

[Meirp]

Hey guys good night from Israel.
I wanted to check with you somethings, as far as i know Veeam doesn't know back up directly to cloud yet (perhaps in next version 12) and I must run the initial backup in 'On-Perm' and then offload will run and move it to cloud.

My question is:
1) how do you backup your VMware to Cloud?
2) In some region i have no local storage so i am on problem, the workaround I found is create azure share or AWS share and map it into Veeam server and run the backup on CIFS it work great but super slowwww, my line bandwidth traffic is 100MB but Veeam works about 2MB
3) This question is general one to my knowledge:

Bear with me, my main VBR server located on Israel as primary Veeam server and I have another Proxy server on Singapore, and I am backing up 2 VM's from Singapore Cluster (VMware), the destination repository is located near to Singapore (CIFS) in Azure, who the hell sending the backup data to Azure? the proxy or the main VBR? (Dedicated job with 2 VM's and one Proxy server selected which also chose on Singapore VMware Cluster)

[HannesK]

Hello,

1) I don't do it myself, but the recommended option today for your size (only very few locations) are having a repository server (Linux box with XFS as file-system is probably the most efficient way) in the cloud if you really don't have any space on-prem
2) yes, that's probably because of the merge operations and expected. Or wrong gateway server
3) it depends how you configured your SMB share. The gateway server is the key component.

Best regards,
Hannes

[Meirp]

what is the suggestion? create the GW on the same location proxy is located?

[Mildur]

For SMB Storage, the Gateway should be the server which has the best and most stable connection to the SMB Storage.

Data Flow is like this:
1) Vmware
2) Veeam Proxy
3) Veeam Gateway Server
4) SMB Storage

If veeam does a synthetic fulls, then all blocks will be copied from „4) SMB Storage“ back to „3) Veeam Gateway Server“, rebuild as a new full backup, and copied again to „4) SMB Storage“.
With slow internet connection, this could be a very time intensive process.
You need to find the shortest path between this components, if the network connection isn‘t fast and stable.

Another option could be to deploy a vm in azure or aws, which connects to the smb share (or better, have disks for backup data attached) I think, that could optimize the backup repo component.

[john_wood]

Hi @Meirp

We have a reference architecture for VMC on AWS - Link

You can deploy an EC2 instance with EBS volumes and use the capacity/archive tier with copy/move modes to minimise EBS storage.

The reference architecture recommends ST1 EBS volumes, but you could use SC1 which is lower performance but 3x cheaper (depending on region)