FusionCompute VMs

Compatibility

Details about supported versions of FusionCompute, please see UltraVR V100R003C10SPC900 Version Mapping.

In the DR Star network scenario, when you create an FusionSphere protected group, only the storage with DR Star is supported (V3 series V500R007C20 and later versions, Dorado series V300R002C00 and later versions).

VMs

-

Non-host-based replication DR

• Public
  • DR can be implemented only for created (based or not based on templates), fully cloned, and template VMs.
  • Protected VM disks support only virtualized SAN storage and Huawei Distributed Block Storage.
  • This DR plan is not applicable to management VMs, linked clone VMs.
  • This DR plan cannot be performed on VMs with peripherals such as GPUs and USBs. If peripherals are added to VMs for which DR has been created, the DR plan cannot be performed on the peripherals.
  • This DR plan is not applicable to VMs using non-persistent disks.
  • VMs using raw device mapping (RDM) shared disks can be used for DR.
  • This DR plan is not applicable to VMs without NICs or VM disks.
  • This DR plan is not applicable to VMs with disks in VHD format on data stores.
  • The HA solution supports DR for VMs where UltraVR is deployed, but does not support DR for the following VMs:

    VMs with SCSI transparent transmission disks and disks in SCSI transparent transmission mode

    VMs for which the Security VM option is enabled

    VMs with snapshots

    VMs with shared disks

  • Only asynchronous replication (centralized storage) in the active/standby DR solution supports DR for VMs that use raw device mapping (RDM) shared disks.
  • Planned migration is not supported when a storage link is faulty.
  • DR is not supported for VMs whose Boot Device is Specify Device Boot Sequence.
  • DR is not supported for VMs whose Security VM Type is SVM.
  • VM-host binding DR is not supported.
  • HPET DR cannot be enabled for VMs.
• FusionCompute
  • This DR plan is applicable to user VMs and VM templates provisioned by the FusionCompute.
  • FusionCompute is required to create datastores using storage resources provided by Fibre channel SAN, IP SAN, Advanced SAN (FusionCompute supports only versions from 5.0 to 6.1), and Huawei Distributed Block Storage. On the datastores, VMs are created. In addition, cluster, host, and port group resources can be provided externally. FusionCompute has interconnection users.
  • FusionCompute 8.0.0 does not support DR protection using Huawei Distributed Block Storage 6.x.
• FusionManager
  • This DR plan is applicable to VMs that belong to the VDC and are provisioned by the FusionManager.
  • This DR plan is not applicable to applications in the FusionManager, non-VDC VMs, and VM templates.
  • FusionManager can be cascaded with FusionCompute to create VMs and provide network and security group resources externally via VPC users.
  • To implement DR for VMs provisioned by the FusionManager, you are advised to configure the network types at the production and DR sites consistently.

    Restrictions for restored VMs differ by network type. For details. see Table 2. Because the IP addresses of VMs may change after a DR switchover, the administrator needs to manually adjust the services on the VMs if the services are associated with IP addresses.

• Huawei Distributed Block Storage
  • Storage replication DR of Huawei Distributed Block Storage with FusionManager is not supported.
  • DR volumes (volumes configured with replication pairs) do not support disk migration.
  • Datastore at the DR site must be created during environment preparation.
  • DR for VM snapshots is not supported (no snapshot is available at the DR site). If a snapshot is created for a DR VM, the snapshot will be deleted with the production VM during DR planned migration and reprotection.
  • Port groups can be accessed only using VLANs.
  • Only synchronous and asynchronous replication networks support shared disk VM DR.

Host-based replication DR (support only versions from FusionCompute 5.0 to FusionCompute 6.1)

• Supported scenarios:

  This DR plan is applicable to VMs using virtual disks or Huawei Distributed Block Storage disks.

• Unsupported scenarios:
  • This DR plan is not applicable to VMs using hybrid disks (virtual and physical disks).
  • This DR plan is not applicable to linked clone VMs, FT VMs, PVS VMs, VM templates, and VMs using shared volumes, pass-through devices (such as USB flash drives, GPUs, and HBAs), remote devices (such as remote USB flash drives and remote CD-ROMs), and raw device mapping (RDM) disks.
  • This DR plan is not applicable to scenarios including VM migration as a whole and online/offline expansion of disks.
  • This DR plan is not applicable to memory volumes on VMs.
  • This DR plan is not applicable to incremental replication when a single disk is added.
  • This DR plan is not applicable to VMs using non-persistent disks.
  • This DR plan is not applicable to VMs without NICs.
  • This DR plan is not applicable to applications in the FusionManager, non-VDC VMs, and VM templates.
  • This DR plan cannot grantee data consistency between dependent VMs or applications. For example, it cannot be ensured that all Oracle RAC VMs are restored to the status at the same time point.
  • This DR plan cannot grantee data consistency between VMs of the same application instance. It cannot be ensured that VMs are restored to the status at the same time point.
• Others:
  • VRG is one of the FusionCompute's component. Use a VRG that meets version requirements in FusionCompute.
  • When a system is switched from the production site to the DR site or switched back from the DR site to the production site (including planned switchover and failover), the existing VM snapshot chain is not switched together with the system. To be more specific, host-based replication-based DR replicates only latest data, excluding the snapshot chain.

Others

• VM specifications are periodically synchronized between sites. When a DR switchover is in progress, the VM specifications at the DR site may be different from those at the production site. Therefore, you are advised to minimize VM specification changes.
• Cache data may fail to be written into disks during LUN remote replication. As a result, data consistency cannot be guaranteed, and VMs may fail to start during the DR process.
• When planning and deploying a DR environment, ensure that datastores in a cluster must be associated with all hosts in the cluster.
• On the DR management page, a storage mapping needs to be added for the target storage to which VM data is migrated.

Static injection

• The production site and the DR site are in the same network segment.

  If no VMs at the DR site are on this network segment, the network of VMs can be completely restored.

• The production site and DR site are in different network segments.

  After VMs are restored, you need to manually configure IP addresses that are reassigned by the system for VMs. To query the reassigned IP addresses, click IP Address Usage in the virtual private cloud (VPC) network to which VMs belong.

Manually configured IP address

• VMs of the production site and the DR site must be planned in different MAC address segments so that the original MAC address of a VM can be retained after a DR switchover. In this manner, the manually configured IP address of the VM will not be lost.
• If a VM's MAC address conflicts, it will be assigned to a new MAC address and its manually configured IP address will be lost.

External DHCP

After VMs are restored, IP addresses are not controlled by the DR system but assigned by an external DHCP server.

Internal DHCP

• The production site and the DR site are in the same network segment.

  If no VMs at the DR site are on this network segment, the network of VMs can be completely restored.

• The production site and DR site are in different network segments.

  IP addresses of the VMs are automatically reassigned.