GitHub user RyanP-Testing added a comment to the discussion: CloudStack-KVM 
integration for Veeam B&R

This is really interesting work, especially because the integration is 
CloudStack-aware rather than treating KVM simply as a generic hypervisor.

One area I’m particularly interested in is the data path and incremental backup 
behavior. For production CloudStack environments, the actual amount of data 
read and transferred during an incremental backup matters just as much as 
whether the operation is technically supported. If the implementation is 
relying on QEMU/libvirt dirty bitmaps or equivalent change tracking, it would 
be useful to understand how that state is maintained across VM reboots, live 
migrations, storage operations, and backup-chain failures.

Storage support is another important point. NFS/local/SharedMountPoint makes 
sense for the initial implementation, but Ceph RBD is quite common in larger 
KVM/CloudStack deployments. Native and efficient handling of RBD would be a 
significant step for broader adoption.

The worker VM approach also looks clean. Using CloudStack to deploy and manage 
the worker keeps the workflow inside the CloudStack control plane, rather than 
introducing another manually managed component. The ConfigDrive/userdata 
requirement is worth highlighting in the documentation, though, as this could 
easily become a deployment gotcha for operators using different network 
offerings.

>From an operational perspective, I’d also be interested in how the integration 
>behaves during less-than-perfect scenarios:

* interrupted full/incremental backup
* failed backup after the dirty bitmap has been updated
* VM migration during a backup
* host failure during backup
* restore to a different cluster/host
* restore when the original network or storage is unavailable
* backup-chain consistency after a failed incremental job

Those scenarios are usually where the difference between a working integration 
and a production-ready integration becomes apparent.

Overall, this is a very positive development for CloudStack/KVM. Having a 
Veeam-native integration can remove a significant barrier for organizations 
considering VMware → CloudStack/KVM migrations, particularly where Veeam is 
already deeply embedded in their backup infrastructure.

Looking forward to seeing the final implementation and the storage/support 
matrix once the remaining QA and Veeam validation are complete.

Ryan P.
Cloud  & DevOps Engineer 
AccuWeb.cloud

GitHub link: 
https://github.com/apache/cloudstack/discussions/12869#discussioncomment-18304544

----
This is an automatically sent email for [email protected].
To unsubscribe, please send an email to: [email protected]

Reply via email to