Limit Max Speed

Limit the network bandwidth for the live migration task to prevent it from using excessive network resources and impacting other services. Scenarios where the parameter takes effect: migrate a VM across nodes with auto selection of run location disabled or migrate across resource pools. Effect and impact: After the parameter is enabled or configured, the migration bandwidth will be set to a specified value. This ensures normal network operation for other services, but the migration time will increase accordingly.

Specify a value from 50 MB/s to the bandwidth value of the physical interface. The parameter should not be enabled for high-bandwidth scenarios (≥10 Gbps). The parameter can be set to 80% of the bandwidth for low-bandwidth scenarios (≤1Gbps).

Max IO Speed

Limit the disk IO speed for the live migration task to prevent it from using excessive disk resources and causing IO latency for other services. Scenarios where the parameter takes effect: Migrate a VM across run locations, storage locations, or resource pools, and the resource pool version is 6.11.1 or later. Effect and impact: After the parameter is enabled or configured, the disk IO speed will be limited during migration. This reduces the impact on the disk IO speed for the service but may prolong the migration time.

Configure the parameter based on the performance of the source Datastore volume. Recommended value for HDD: 200 MB/s, recommended value for SSD: 500 MB/s. The value cannot be set too low, otherwise, migration timeout may occur.

Compressed Transmission

After the parameter is enabled, the migration data will be compressed by using the self-developed Sangfor4 algorithm, thereby improving migration speed. This process may consume up to 4 CPU threads. Scenarios where the parameter takes effect: All live migration scenarios. Effect and impact: After the parameter is enabled or configured, the migration rate will be increased by over 7 times, and the CPU consumption will be reduced by 80%. However, it will use some CPU resources on the source node. If the CPU on the source node is full loaded, the service may be impacted.

The parameter should be enabled for low-bandwidth scenarios (≤1 Gbps) and disabled for high-bandwidth scenarios (≥10 Gbps). Monitor the CPU usage on the source node during migration to prevent it from exceeding 80%.

CPU Throttling

CPU Throttling: Limit the VM's CPU performance to control the generation rate of dirty data, preventing migration failure caused by data being generated too quickly. Scenarios where the parameter takes effect: Migrate a VM during peak hours. Effect and impact: After the parameter is enabled or configured, the VM's CPU performance will be limited during migration. This reduces the generation rate of dirty data and increases the migration success rate. However, the VM's business performance will be temporarily degraded.

Enable the parameter only if the migration task fails multiple times. Configure the CPU performance limit rate to no less than 50%. Disable CPU throttling promptly after the migration completes.