What Is Change Block Tracking? How CBT Speeds Up Backups and Protects VMs
Change block tracking is the quiet technology that makes modern virtual machine backups fast enough to run every day. Without it, a backup system would have to read every block of every virtual disk, every time, just to find what had changed. With it, the hypervisor keeps a record of changed blocks, so the next backup can copy only new or modified data. That difference matters for speed, storage, network load, and recovery time, especially when a virtual machine holds terabytes of data.
This guide explains what change block tracking is, how it works, how it differs across platforms, and how to use it safely. It also includes a worked example of the time and data it can save on a large virtual machine, plus a troubleshooting checklist for the common case in which tracking fails or resets.
What Is Change Block Tracking?
Change block tracking, often shortened to CBT, is a hypervisor feature that records which blocks of a virtual disk have changed since the last backup or snapshot. A block is a small, fixed-size unit of storage. When an application writes to a virtual machine, the hypervisor marks the affected blocks as changed. Backup software can then ask for that list, rather than scanning the whole disk.
Think of a large book in which only a few pages are edited each day. Without an index of edited pages, a librarian would have to reread the entire book to find the changes. Change block tracking is that index. It does not copy the data itself. It simply tells the backup tool where to look, which keeps incremental backups small and quick.
How Change Block Tracking Works
Change block tracking works through a cycle that repeats with each backup. First, the backup software takes a snapshot of the virtual machine, creating a stable point in time to copy from while the live machine keeps running. Next, it asks the hypervisor for the list of blocks changed since the previous backup. It then reads only those blocks and stores them in the backup repository. Finally, it records a new reference point for the next job.
On the first backup after change block tracking is enabled, a full backup is normally required, because there is no useful change list yet. That full backup becomes the baseline, and later incremental backups build on it by copying only changed blocks. If the tracking file is deleted, corrupted, moved, or reset, the backup software can no longer trust the change list, and it should fall back to a full backup.
Change Block Tracking Across Hypervisors
The concept is shared, but implementation details matter for backups, migrations, and restores. Always check your backup vendor’s support matrix, because a hypervisor feature alone is not enough: the backup software must also know how to query it.
| Platform | Tracking name | How it is enabled | Key point to remember |
|---|---|---|---|
| VMware vSphere | Change Block Tracking (CBT) | Per virtual machine, then a full backup to create a baseline | Storage migration and snapshot consolidation can reset CBT |
| Microsoft Hyper-V | Resilient Change Tracking (RCT) | Built into modern Hyper-V virtual machines and checkpoints | RCT is designed to survive many operations that reset older methods |
| XenServer / Citrix Hypervisor | Changed block tracking | Enabled for the virtual machine, used through the platform API | Support depends on the backup vendor and virtual machine version |
| KVM and QEMU | Dirty bitmaps | Managed by the backup or snapshot tool using QEMU bitmaps | Bitmaps must be handled carefully across failed jobs, or a full scan is safer |
| Nutanix AHV | Changed region tracking | Integrated with the platform and its backup APIs | Native snapshots and replication often work alongside third-party tools |
For readers who run security virtual appliances, the same principle applies: the appliance is itself a virtual machine, so its disks benefit from change block tracking when it is backed up.
Why Change Block Tracking Matters for Backups and Security
The most obvious benefit is speed. Shorter backup windows mean less contention with production workloads, fewer jobs that overrun into business hours, and a better chance of meeting recovery point objectives. When backups finish quickly, you can also run them more often, so less work is lost if a virtual machine fails or is encrypted by ransomware. Copying only changed blocks also reduces data written to the backup repository and moved across the network, which lowers cloud storage costs. The facilities that house this infrastructure matter too, as explained in this overview of what data centers are and how they work.
Worked Example: Backup Time and Data Saved on a Large Virtual Machine
Consider a file and database virtual machine with a 4 terabyte virtual disk. Suppose only 2 percent of its data changes on a typical day. That is about 80 gigabytes of changed blocks, compared with 4,096 gigabytes for a full copy. The difference changes whether a backup window is practical at all.
At a sustained transfer rate of 500 megabytes per second, copying the full 4,096 gigabytes would take about 8,192 seconds, or roughly 2 hours and 17 minutes, before compression, deduplication, and verification. Copying 80 gigabytes at the same rate would take about 160 seconds, or under 3 minutes of raw transfer time. Even after allowing 15 minutes for snapshots, cataloguing, and verification, the incremental job could finish in roughly 18 minutes instead of more than 2 hours, saving about 87 percent of the time and about 98 percent of the copied data. These figures are illustrative, not a promise: real results depend on block size, change rate, compression, and storage speed.
CBT Failure and Reset: Troubleshooting Checklist
When change block tracking fails, the safest response is usually a new full backup, not a forced incremental. Use this checklist in order, and record what you changed so the next job can be audited. Protect the backup system itself as carefully as production: use separate credentials, multi-factor authentication, and network segmentation, following the risk controls described in our guide to private cloud security.
- Check the backup job log for the exact error, including VMware error 4292, a CBT reset warning, or a message that the change tracking file is missing or inconsistent.
- Confirm that CBT is still enabled and that no recent storage migration, snapshot consolidation, disk move, or restore has reset it.
- Look for failed or interrupted jobs, power loss, or storage outages that may have left a snapshot or bitmap in an inconsistent state.
- Consolidate or remove orphaned snapshots, then run a new full backup to create a clean baseline before returning to incremental jobs.
- Verify that the backup software, hypervisor, and any backup proxy are on supported versions, and update them together rather than one component at a time.
- Test the restore, not just the backup. Restore to an isolated network and check that files, databases, and services start correctly.
Do not simply clear the error and continue with incrementals if you are unsure whether the change list is complete. A backup that looks successful but misses changed blocks is worse than a slow full backup, because the gap may only appear during a restore.
A note on scope: this guide focuses on virtual machine backups, not on file-level change tracking inside an operating system, although the underlying idea of recording changed blocks is similar in both cases.
Conclusion
Change block tracking turns an otherwise expensive, slow backup process into a practical daily control. By recording only the blocks that changed, it reduces backup time, storage, and network load, and it makes frequent recovery points possible for large virtual machines. Its value depends on correct setup, regular full baselines, careful handling of resets, and tested restores.
Frequently Asked Questions
What is change block tracking?
Change block tracking is a hypervisor feature that records which blocks of a virtual disk have changed since the last backup, so backup software can copy only those blocks instead of scanning the whole disk.
Does change block tracking replace full backups?
No. It depends on a full backup as a baseline, which must be combined with incrementals during a restore.
Why does change block tracking sometimes reset?
It can reset after storage migration, snapshot consolidation, a disk move, a restore, a failed job, or a change that invalidates the tracking file. Then run a new full backup.
Which hypervisors support change block tracking?
VMware vSphere supports CBT, Microsoft Hyper-V uses Resilient Change Tracking, and XenServer, KVM and QEMU, and Nutanix AHV offer their own tracking methods.
How much time and data can change block tracking save?
The saving depends on the change rate. In the example above, a 4 terabyte machine changing by 2 percent per day copied 80 gigabytes instead of 4,096 gigabytes.


