This solution is applicable to local and remote DR. The production center and DR center work in active-passive mode. If a disaster occurs, the DR center quickly takes over services from the production center, ensuring optimal service continuity. The replication relationship is established between storage systems used by host services for remote DR. In addition, this solution can protect HyperVault snapshots of protected objects for rolling back data, preventing data loss and data corruption caused by viruses.
Context
As IT services being increasingly used, continuous and secure running of IT systems becomes critical for normal enterprise operation. IT systems in use face many problems:
- Device faults: IT systems cannot work properly. Replacing devices may interrupt services and result in a long recovery period.
- Regional outages: The entire IT system cannot work, interrupting all services.
- Fires occurred on the data center. All IT systems are damaged, data loose, and all services are interrupted.
- Natural disasters such as fire, earthquake, and mudflow may cause imperative damages on IT systems, interrupting all services and leaving no data.
To enhance the IT system availability to address the preceding problems, Huawei launched the active-passive DR solution.
Solution Description
This solution enhances IT system availability by establishing a same-city or remote DR center. If a disaster occurs on the production center, the DR center quickly takes over production services to minimize service download, thereby reducing customer loss. The underlying storage interworks with the upper-layer virtualization platform to implement FusionCompute VM DR.
The following lists the supported DR technologies:
- Synchronous replication (SAN)
- Asynchronous replication (SAN)
- Host replication
Storage replication is applied in synchronous replication (SAN), asynchronous replication (SAN).
Deployment Modes
Centralized and distributed deployment modes are supported:
- Centralized deployment: One UltraVR is deployed on a server at the DR center to provide DR management for production and DR resources. The UltraVR must communicate with the production center properly over networks.
- Distributed deployment: Two UltraVRs are respectively deployed on the servers at the production and DR centers. Each provides DR management for resources of the owning center. The network between the two UltraVR must be connected properly.
Compared with centralized deployment, distributed deployment is more secure and provides shorter latency during daily production center maintenance. Up to two UltraVRs can be deployed.
Both local and remote DR modes are supported.
- Local DR: The production and DR centers are located in the same data center and deployed in centralized mode.
- Remote DR: The production and DR centers are located in different data centers and can be deployed in either centralized or distributed mode.
Array-Based Replication DR
DR solutions using this technology can implement minute-level RPO and hour-level RTO. The specific RPO and RTO vary with site conditions.
This technology utilizes the remote replication function provided by storage systems to replicate VM data from the production center to the DR center, to replicate and protect data of the production center. For details about the implementation principles of array-based replication, see the Array-Based Replication DR Principles.
Application scenarios:
- Huawei SAN is used and production and DR centers are connected over an IP network.
- The distance between production and DR centers is flexible.
- VMs are migrated across different centers as planned.
- Service continuity must be ensured.
- VMs are restored according to their priorities or dependencies.
- DR tests have higher priorities.
The typical DR networks in distributed mode are shown in the following figures.
- Synchronous replication (SAN) DR and asynchronous replication (SAN) DR are same in network, though their implementation principles are different. Figure 1 shows the network.
Host-Based Replication DR
If the host-based replication DR technology is required, ensure that the FusionCompute version is between 5.0 and 6.1.
The host-based replication DR used by Production application host VMs provides RPO in seconds and RTO in minutes. The specific RPO and RTO vary with site conditions.
Host replication DR uses the host-based replication function of to remotely replicate VM data from storage devices at the production center to the DR center. Host replication DR also uses Production application host to replicate VM configuration data (including VM CPU, memory, network adapter, and disk attributes) and manage DR plans. For details about its working principles, see Host-Based Replication DR Principles.
The host-based replication DR applies to non-key services of small and medium-sized enterprises (SMEs), such as Enterprise Resource Planning (ERP), email servers, and desktop clouds.
Figure 2 shows the network of host-based replication (SAN) DR in distributed mode.
DR Management
Production application host provides the following functions:
- Automatically aware VMs running virtualization infrastructure (FusionManager or FusionCompute) being added.
- Protect service FusionCompute VMs based on preset DR policies.
- Support one-click tests to verify data availability and estimate the recovery time objective (RTO) in routine O&M.
- If a planned disaster (such as an outage or system maintenance) is to happen:
- Migrate services from the production center to the DR center through one-click scheduled migration.
- Use the re-protection function to protect data at the DR center after the service switchover is complete.
- If an unplanned disaster (such as an earthquake, fire, or flood) occurs, restore services by using backup data in the remote DR center by one click.