Continuous Data Protection has been one of Veeam’s most powerful — and most narrowly scoped — features for years. Until 13.1, CDP was exclusively a VMware vSphere capability. That changes with Universal CDP, which extends near-zero RPO protection to a broader set of workloads.

This is a genuinely significant architectural shift. Let’s work through what it means, what the current limitations are, and how to incorporate it into your DR strategy realistically.

What is Continuous Data Protection?

Before covering what’s new, it’s worth being precise about what CDP actually is — because the term gets misused.

Traditional backup and replication work on a scheduled basis. You run a backup job every hour, or a replication job every 15 minutes. In a failure scenario, your recovery point is the last completed job. If that job ran 55 minutes ago, you potentially lose 55 minutes of data. That gap is your Recovery Point Objective (RPO).

CDP takes a fundamentally different approach. Rather than periodic snapshots, CDP captures every write to your protected workload in near real-time, continuously streaming changes to the replica. In a failure scenario, your RPO is measured in seconds — not minutes or hours.

For workloads where data loss is genuinely unacceptable — databases, financial transaction systems, healthcare record systems — CDP is the right tool. For everything else, scheduled backup and replication remain appropriate.

What Universal CDP Adds in 13.1

Prior to 13.1, Veeam’s CDP was built specifically around VMware’s I/O filter technology (vSphere APIs for I/O Filtering, or VAIO). This tight VMware integration is what made it work, but it also meant CDP was unavailable for any non-VMware workload.

Universal CDP in 13.1 breaks that dependency. By moving the I/O capture mechanism away from a hypervisor-specific API, Veeam can extend CDP protection to workloads running on other platforms.

The core protection model remains the same:

• Continuous write capture — every I/O operation on the protected workload is captured as it happens

• Near-real-time replication — captured changes stream continuously to the CDP replica

• Journal-based point-in-time recovery — the CDP journal allows you to recover to any point within the journal retention window, not just the last replication checkpoint

• Fast failover — the replica is always current, so failover is measured in seconds to low minutes

The Journal: Your Flexible Recovery Window

One of CDP’s most underappreciated capabilities is the recovery journal. Unlike traditional replication, which gives you replica states at discrete replication intervals, the CDP journal captures every write continuously. This means you can recover to any point in time within the journal window — not just “the replica as of 14:00.”

This matters enormously for ransomware scenarios. If ransomware begins encrypting data at 14:23 and you do not detect it until 15:15, you can recover to 14:22 — a moment before the encryption began — rather than being forced to take the 14:00 replica checkpoint that may already contain corrupted data.

Set your journal retention window to give yourself enough lookback to catch threats before they become unrecoverable. A minimum of 24 hours is a reasonable baseline; longer for environments with slower threat detection capabilities.

Current Limitations — Read These Before You Plan

I want to be direct about this section because I have seen organizations design DR runbooks around features before checking whether those features are fully implemented. Universal CDP in 13.1 has known limitations that affect how you can use it:

Planned failover is not yet supported. In a traditional replication scenario, planned failover means gracefully shutting down the production workload, allowing the replica to sync fully, then bringing up the replica as the new primary. This is the clean path for scheduled maintenance or site migrations. Universal CDP in 13.1 does not yet support this workflow.

Failback to the original location is not yet supported. After failing over to a CDP replica, getting back to your original location requires a manual process rather than a Veeam-orchestrated failback. This extends your recovery timeline and introduces operational complexity in a scenario where you want things to be as straightforward as possible.

These are significant gaps for production DR planning. They do not make Universal CDP unusable — emergency failover works, which is the primary scenario CDP addresses — but they do mean you should plan your DR runbooks carefully around the manual steps required after failover.

Veeam’s track record on closing feature gaps across releases is generally good, and I expect these limitations to be addressed in subsequent releases. But plan for what exists today, not what might ship tomorrow.

Workloads Right for Universal CDP

Given the limitations above, which workloads should you actually protect with Universal CDP in 13.1?

Strong candidates:

• Databases — transaction logs make databases highly sensitive to data loss, and CDP’s near-zero RPO aligns with that sensitivity

• Financial processing systems — where data loss has direct regulatory and financial consequences

• Healthcare systems — patient data systems where data integrity and continuity are compliance requirements

• Any workload where your business has defined a sub-minute RPO

Approach with caution (for now):

• Workloads requiring frequent planned failover — until planned failover support arrives, Universal CDP adds friction to scheduled maintenance

• Environments with complex failback requirements — the manual failback process increases risk during recovery

For workloads that do not require near-zero RPO, scheduled replication with a tight interval (15-30 minutes) remains a more operationally complete solution while Universal CDP matures.

CDP vs. Replication: Choosing the Right Tool

Since Veeam now offers both Universal CDP and native replication for an expanding set of platforms (including Proxmox in 13.1), the question of which to use is increasingly relevant.

The answer is not one or the other for your entire environment — it is applying the right tool to each workload tier based on your actual RPO requirements. Most workloads do not need sub-minute RPO. The ones that do are worth the additional complexity CDP brings.

Configuring Universal CDP in Veeam 13.1

The configuration workflow for Universal CDP follows Veeam’s standard policy-driven approach:

1. Add the source hosts and workloads to Veeam inventory

2. Create a CDP Policy rather than a standard replication job

3. Select the protected workloads

4. Define the target (replica destination host and storage)

5. Configure the RPO target — how frequently you want the CDP engine to write checkpoints to the replica

6. Set the journal retention — how far back you want point-in-time recovery to extend

7. Enable the policy and monitor CDP health from the console

Monitor your CDP policy health actively. CDP is more sensitive to network and storage performance than scheduled replication — sustained throughput matters because changes are flowing continuously rather than in batches.

Planning for the Feature Gaps

Until planned failover and automated failback arrive for Universal CDP, build the following into your DR documentation:

Manual failback procedure: Document the steps to migrate workloads back to the original location after a CDP failover. This should include data synchronization back to the source, validation, and cutover. Test this procedure at least quarterly.

Maintenance window planning: For workloads on CDP that require maintenance involving source downtime, plan for the additional manual coordination steps that planned failover would otherwise handle.

RPO validation: Because CDP is continuous, the RPO you achieve depends on the health of the CDP pipeline — network bandwidth, storage latency, and system load all affect it. Monitor your actual achieved RPO against your target, not just your configured checkpoint interval.

Bottom Line

Universal CDP in Veeam 13.1 is a genuinely important capability expansion — extending near-zero RPO protection beyond VMware is a meaningful step for organizations running diverse workload platforms. The current limitations around planned failover and failback are real and require careful DR planning to work around.

For your highest-criticality workloads where sub-minute data loss is unacceptable, Universal CDP is the right technology. Deploy it with clear eyes about what it currently does and does not do, build your runbooks around the actual capabilities, and watch for the feature gaps to close in subsequent releases.

As always, test your failover and recovery procedures in a non-production environment before you need them in anger.