The disk passthrough feature allows you to directly connect the local physical disks (in Non-RAID or JBOD mode) of HCI nodes or external LUNs (using iSCSI or FC protocol) to VMs. VMs with disk passthrough enabled are ideal for special business scenarios where direct access to physical disks is required, such as high-performance LUN mapping of external SAN storage, data recovery services (directly reading raw data from physical disks), and the deployment of industrial software that relies on independent physical disk drivers.
Common Features of VMs with Disk Passthrough Enabled
Feature
Supported or Not
Offline Specification Change
Yes. You can change CPU and memory specifications. To change a physical disk associated with the VM, disassociate the physical disk first.
Online Specification Change
Yes. You can only increase the number of CPUs and the memory size. To mount or unmount a physical disk, power off the VM first.
Cold Migration (Run Location)
No. VMs associated with the physical disk of the node through disk passthrough can only be migrated along with the node (cross-node migration is not supported).
Live Migration (Run Location)
No. Live migration is not supported for VMs associated with physical disks through disk passthrough.
HA
No. VMs cannot be automatically migrated to another node when the node where they currently reside fails. You need to manually migrate the VMs and associate them with a spare physical disk.
VM Snapshot
No. The VM snapshot does not contain data in the physical disk. To back up data in the physical disk, use an external backup tool.
Cloning
No. The physical disk cannot be cloned. You need to create a disk image before cloning the VM.
Disk Provisioning Type Selection
No. Dynamically and thinly provisioned disks are not supported. Only pre-allocated disks are supported (the physical disk capacity is fixed).
Space Reclamation
No. You can reclaim the space of a physical disk associated with a VM through disk passthrough only on the storage server.