Veeam Agent Windows Restore Limit

Hi ​@christiandeuerlein 

Can you provide more information about the activity you are trying to perform?
Are you trying to restore a VM from an Azure blob to your on-premises infrastructure? Can you provide more details?
What type of storage is the restore target (local SSD/HDD, VM disk, physical server)? And are you seeing high CPU usage during the restore?

If so, what type of connectivity do you have from your on-premises infrastructure to Azure? VPN STS\Express route?

Thank you.

Also what Veeam versions as well helps for suggestions.

Hi, it is a full disk restore, the backup is stored in a Azure Blob storage.

The VM onPrem runs on Hyper-V, on current hardware (Intel 6527P) and highend SAN. CPU usage is about 2 - 5% during restore (8 vCPUs/16 GB VM).

Connectivity is > 10 GB, no express route.

Disk writes 100 MB/s at 0-1 queue length. There is > 500 MB/s available. 

I have to check that, should be the latest.

Probably not, we are using: https://download2.veeam.com/VAW/v6/VeeamAgentWindows_6.3.2.1302.zip

The latest one is v13 which you may want to try possibly not saying that is the answer but a good test.

​@christiandeuerlein Would it be possible to validate if everything is OK on the source side? Could you somehow test if the source is intact, try validating if the source has any corrupted blocks, perhaps using a tool from Azure itself?

The sources are fine. We are using Veeam Agent to transfer different customer data to our own datacenter. Throughput is always 100 MB/s (Disk, Network varies).

When you put disk thoughput counter in performance monitor, it is a straight line. Strange.

Have you created a support ticket to see what they say?  I think that might be the better route now to take.

Is there a firewall between Azure and Hypev?

Firewalls tend to limit bandwidth.
Standalone Veeam Agent for Windows (free or paid) has a configuration for throttling the bandwidth.

Processing bottleneck - Veeam decompresses and processes data before writing. If CPU is saturated, it limits how fast data can be prepared for disk.

Block size settings - Veeam's restore block size might not be optimal for your storage configuration.
Cache repository settings - If using a cache repository during restore, its performance affects the overall speed.

regards

In my opinion, it’s likely a storage account limit: https://learn.microsoft.com/en-us/azure/storage/common/scalability-targets-standard-account

Azure doesn’t charge per API call like AWS does, but it does limit calls per second, and this is relevant because Veeam doesn’t store monolithic files, it chunks out the datablocks into individual blobs; a rough estimate is 1000000 blobs per 1 TB of data when the backup uses default 1 MB block size (1 TB / 1 MB = 1048576 blobs for a 1 TB file)