I85PREP
⌘ K
HCIE-Cloud study materials
HCIE-Cloud study resources

HCIE-Cloud: Cloud Platform

HCIE-Cloud · Work through your exam, one topic at a time.

All guide sections

Cloud Platform

This domain tests how well you understand the architecture, deployment, and day-2 operations of the Huawei Cloud Stack platform. Expect questions on FusionCompute components, resource pooling, high availability, multi-tenancy, and the management plane rather than on individual cloud service features.

Core architecture

  • FusionCompute splits into CNA on each compute node and VRM as the cluster manager. VRM runs active/standby; losing one VRM does not stop running VMs.
  • A cluster groups hosts with compatible CPU features and shared storage access. DRS balances workload, while HA restarts VMs after a host failure.
  • ManageOne provides O&M and service provisioning, and ServiceOM handles resource operations. Keep management-plane and service-plane networks separate.
  • VDC and quota models provide tenant isolation; resource pools must reserve capacity for HA and management overhead.

Worked example

A tenant reports that a VM did not restart after its CNA host failed. The cluster HA switch is on, but the cluster has only one remaining host with 90% CPU and memory allocated. Reasoning: HA restart needs reserved capacity or an admission-control policy that leaves headroom. Check the cluster HA policy, host maintenance state, and whether anti-affinity rules prevent placement on the surviving host. Also verify VRM health and that the VM is not excluded from HA. The fix is to add capacity or lower the HA reservation, not to enable DRS.

Traps that cost marks

  • Confusing VRM HA with VM HA: VRM HA protects the management component, not tenant VMs.
  • Assuming DRS migrates VMs after a host crash. DRS is for balancing; HA performs restart.
  • Ignoring cluster CPU compatibility, which blocks live migration and HA placement.
  • Treating a host in maintenance mode as available for HA restarts.

Key points

  • CNA runs on hosts; VRM is the active/standby cluster manager.
  • DRS balances, HA restarts, and admission control reserves capacity.
  • VDC and quotas enforce multi-tenant isolation.
  • Management and service networks should be separated.