Problem Description
We are operating a 2-node Windows Server 2022 Hyper-V failover cluster, which hosts highly available (HA) virtual machines (VMs). Among these VMs are a Debian 12 VM (48GB RAM) and a Windows Server 2022 VM (4GB RAM).
Issue
Although failover functions smoothly for both VMs, the live migration duration differs significantly.
Windows VM (4GB RAM): The live migration process completes within 1-2 minutes.
Debian VM (48GB RAM): The live migration process takes 5-6 minutes despite similar network and storage configurations.
Technical Details
Hyper-V Hosts: Running Windows Server 2022 Enterprise (fully patched).
Cluster: Comprised of 2 nodes with shared storage (Cluster Shared Volumes, CSV).
VM Configurations:
Debian 12 VM:
- 48GB RAM
- 24 vCPUs
- Hyper-V Generation 2 VM
- Static memory allocation
- VHDX stored on CSV
Windows Server 2022 VM:
- 4GB RAM
- 1 vCPU
- Hyper-V Generation 1 VM
- Static memory allocation
- VHDX stored on CSV
- Networking: Dedicated 1 Gbps NIC for live migration
My Observation
After some research, I have found that the longer live migration time for the Debian 12 VM compared to the Windows Server VM is expected due to its significantly larger memory allocation (48GB vs. 4GB).
Key Factors Causing Slow Live Migration
- Memory Size : Live migration transfers RAM contents. A 48GB VM inherently requires more data transfer than a 4GB VM.
- Memory Activity : If the Debian VM is under a heavy workload, Hyper-V must track and transfer memory changes during migration (increased "dirty pages").
I would like to verify if my understanding regarding this is correct.