{{ $t('productDocDetail.guideClickSwitch') }}
{{ $t('productDocDetail.know') }}
{{ $t('productDocDetail.dontRemind') }}
6.11.3
{{sendMatomoQuery("Sangfor Cloud Platform (SCP)","Agentless Backup")}}

Agentless Backup

{{ $t('productDocDetail.updateTime') }}: 2026-04-29

Agentless backup refers to VM backup via third-party backup apps without installing plugins on SCP or VMs. Specifically, by leveraging SCP and HCI OpenAPIs and the disk data access SDK, third-party backup apps can configure and manage data for VMs and their disks to implement VM backup and recovery.

Agentless backup involves the following three main concepts:

  1. In-memory bitmap: When Changed Block Tracking (CBT) is enabled on a VM for the first time, an in-memory bitmap is initialized. Each 1-bit pixel of the bitmap records data changes in a data block of the VM. The pixel value 0 indicates that data remains unchanged, while the pixel value 1 indicates that data has changed. The in-memory bitmap persists in the bitmap file, and all pixel values are reset to 0 after VM backup is complete.
  2. CBT file: When CBT is enabled for a VM, the system creates a CBT file in the datastore where the VM disk resides to record data changes in data blocks. Unlike an in-memory bitmap, a CBT file records the data block changes at different snapshot points, and these data block changes are distinguished by CBT version numbers. The CBT file is persistent and can be incrementally modified.
  3. CBT version number: A CBT version number is a 4-byte serial number used to record the changes of a data block in the CBT file, and it changes with the CBT file number. CBT version numbers persist in the CBT file.

Figure 11 shows the principle of agentless backup. The full backup process and incremental backup process are as follows:

Full backup (t0) process:

Step 1.Take a snapshot of an HCI VM to record the data at time t0, and create and initialize bitmaps and CBT files for all VM disks.

Step 2.Obtain bitmap information of valid data blocks on VM disks from CBT files.

Step 3.Read valid data blocks of the disks from the snapshot taken in Step 1 according to the bitmap information and store them in full images in the backup repository. Record the CBT version numbers of disk data blocks in the snapshot for use in the next backup.

Step 4.After the backup is complete, delete the snapshot taken in Step 1.

C:/Users/00691/Desktop/企业微信截图_17536841413778.png企业微信截图_17536841413778

Figure 11 Principle of Agentless Backup

Incremental backup (t1) process:

Step 1.Take a snapshot of the HCI VM to record disk data at time t1. HCI will update CBT files according to the bitmaps at time t1. After the CBT files are updated, all pixel values of the in-memory bitmaps are reset to 0.

Step 2.Obtain bitmap information of incremental data blocks of VM disks at time t1 from CBT files.

Step 3.Read the incremental data blocks of disks from the snapshot taken in Step 1 according to the bitmap information obtained in Step 2 and store them in incremental images in the backup repository.

Step 4.After the backup is complete, delete the snapshot taken in Step 1 and reset all pixel values of the in-memory bitmaps at time t0 to 0.

Agentless backup is implemented based on external snapshots. Figure 12 shows the disk changes during backup using external snapshots. Under normal circumstances, each VM disk can be represented by a single image file regarded as the original disk. During agentless backup, while a snapshot is taken for the VM, the original disk becomes read-only, and a new image file, namely the incremental disk, is created based on the original disk. The data on the original disk will be written to the incremental disk. This way, data backed up to the snapshot can be obtained through the original disk. When the backup is complete, the snapshot will be deleted, and the original and incremental disks will be merged into a complete disk.

C:/Users/00691/Desktop/企业微信截图_17536843027804.png企业微信截图_17536843027804

Figure 12 Disk Changes During Backup Using External Snapshots