API Doc
Search Docs...
⌘ K
OverviewDeploymentManagementOperationReferenceGlossary

Operating method

DRS can operate in the following two ways:

  • Automatic operation (by default)

    DRS runs automatically at regular intervals of 10 minutes.

  • Manual operation

    If there are no virtual machines being migrated in a cluster, do the following to manually run DRS for that cluster: Click the cluster to access its page, then select All > DRS from the tab bar, and click Generate now.

Operational principles

Each time DRS runs, the system re-evaluates and adjusts virtual machine placement based on VM placement group policies and the resource balance status of the cluster. It first corrects any VM placements that violate the policies, and then balances the cluster load without breaching the placement policies.

  • VM placement group policy

    When a Running virtual machine in the cluster does not comply with its associated VM placement group policy, the system generates a migration recommendation and relocates the virtual machine to a suitable host.

  • Cluster resource balance status

    The system evaluates the balance status of cluster resources by calculating the DRS score for each virtual machine, host, and the entire cluster based on CPU, memory, and storage usage of Running virtual machines.

    If the cluster is determined to be imbalanced, the system generates one or more migration recommendations based on the VM placement group policies and the configured migration threshold. These recommendations are then executed according to the defined automation level, migrating resource-constrained virtual machines to appropriate hosts without violating placement group policies.

    • VM DRS score

      The DRS score of a virtual machine is calculated based on its CPU, memory, and storage usage. The higher the DRS score, the more sufficient the virtual machine's resources. The lower the score, the more the virtual machine contends for resources.

      • No contention: The VM DRS score is greater than or equal to 90%.
      • Mild contention: The VM DRS score is between 80% and 90%.
      • Severe contention: The VM DRS score is less than 80%.
    • Cluster DRS score

      The cluster's DRS score is the average DRS score of all running virtual machines in the cluster and is used to evaluate the overall resource balance status of the cluster. A higher DRS score indicates more balanced resource usage in the cluster.

      • When the cluster DRS score is greater than or equal to 90%, it indicates that the cluster resources are balanced and no optimization is needed.

      • When the cluster DRS score is between 80% and 90%, it indicates a slight imbalance in cluster resources, and you should pay continuous attention to the cluster resource load.

      • When the cluster DRS score is less than 80%, the cluster resources are imbalanced, and you should perform virtual machine scheduling based on migration recommendations to optimize the cluster resources.

      Information:

      If the DRS score of a cluster has been low for an extended period and no migration recommendations can be generated even after you adjust the migration threshold, it might indicate that resource optimization through virtual machine migration is not possible for this cluster. In such cases, you might consider expanding the cluster to provide more sufficient resources.

    • Host DRS Score

      The host DRS score is calculated based on its CPU and memory usage. The higher the DRS score, the more sufficient the host's resources. The lower the score, the more insufficient the host's resources.

      • Sufficient: The host DRS score is greater than or equal to 30%.
      • Mildly insufficient: The host DRS score is between 10% and 30%.
      • Severely insufficient: The host DRS score is less than 10%.