Feature Request Overview 

Large-scale OpenStack deployments running enterprise backup and disaster recovery solutions (e.g., Commvault, DRaaS providers) are increasingly constrained by the lack of a backend-agnostic, VM-level Change Block Tracking (CBT) capability in OpenStack Compute.

While some storage backends may offer array-specific CBT or replication, customers operating heterogeneous or externally managed storage systems require a compute-side mechanism to identify changed blocks at the virtual disk level efficiently. This capability is widely relied upon in VMware-based ecosystems to enable efficient incremental backup and near-continuous replication with low RPO.

This RFE proposes investigating and defining a Nova-orchestrated CBT capability, leveraging existing primitives in QEMU and libvirt, to provide a consumable, vendor-neutral foundation for backup and DR integrations.

Business justification 

This feature would enable customers to:

• Perform efficient incremental backups without requiring full volume scans

• Reduce backup windows and backend read I/O for large VM disks

• Support near-continuous DR workflows at the VM level with lower RPO

• Integrate enterprise backup and DR solutions consistently across heterogeneous storage environments

• Achieve parity with expectations set by other virtualization platforms when migrating large-scale workloads to OpenStack

Without a compute-level CBT capability, customers face scalability and operational challenges that limit OpenStack adoption for enterprise backup and DR use cases.

Functional requirements 

The outcome of this feature should be:

1. A documented design describing how Nova can orchestrate VM-level CBT using existing hypervisor capabilities.
2. A clear definition of:

• • How CBT can be enabled, tracked, and managed across the VM lifecycle.
  • How CBT state aligns with instance operations such as start, stop, snapshot, migrate, and delete.

      3. Identification of constraints, limitations, and failure scenarios relevant to backup and DR integrations.

      4. An architectural analysis of Nova, libvirt, and QEMU interactions required to support CBT in a safe and scalable manner.

      5. Alignment with upstream OpenStack architectural principles and service boundaries.

Describe the customer impact

In the absence of compute-level CBT:

• Incremental backups rely on full disk reads and CRC/hash-based comparison.

• Backup windows grow linearly with disk size.

• Storage systems experience high read amplification during backup operations.

• Near-continuous DR at the VM level becomes impractical at scale.

This directly impacts customers operating large OpenStack environments who require predictable backup windows, low RPOs, and efficient DR workflows.

(Optional) Point of contact

• Provide any additional points of contact for this feature request, such as an account executive, SA, or TAM: 
•  

(Optional) Additional links

Click More > Link to add any links to issues, such as an outcome, that are related to this feature request.