Depending on the power status of virtual machines during migration, there are three migration methods: hot migration, cold migration, and staged migration.
In a hot migration, virtual machines remain started throughout the migration process and do not need to be shut down. Businesses running on virtual machines will not be interrupted during migration, but the target host must be compatible with the CPU of the virtual machines.
During migration, the established VNC (Virtual Network Computing) connection between the browser and the virtual machines will be interrupted. To restore the connection, you need to refresh the VNC interface. Note that this VNC connection interruption does not affect the virtual machines' operation.
Hot migration is applicable to intra-cluster migration and cross-cluster migration where the target cluster is an ACOS cluster.
In a cold migration, virtual machines are migrated in a shutdown state. Cold migration has lower requirements on the cluster networks and the CPU compatibility between the target host and the virtual machines.
Cold migration is applicable to intra-cluster migration or cross-cluster migration where the target cluster is an ACOS cluster.
In the first phase of a staged migration, virtual machines will be migrated in a started state, and you need to manually shut them down in the second phase. After the migration is complete, the virtual machines will automatically restart. This method has lower requirements on the cluster networks and the CPU compatibility between the target host and the virtual machines; however, the virtual machines must have an operating system installed.
During a staged migration, virtual machines will initially remain started, and the system will migrate all their current data. When the migration is nearly complete, you need to manually shut down the virtual machines so that the system can continue to transfer any incremental data generated by the started virtual machines during the migration. Once the migration is complete, the virtual machines will automatically restart.
Since the downtime of the virtual machines during a staged migration is relatively short and has little impact on the businesses running on it, this migration method is generally used in scenarios where there is no direct network connection between the source cluster and the target cluster and where a short downtime is required.
Staged migration is only applicable to cross-cluster migration where the target cluster is an ACOS cluster.
