Hybrid Disaster Recovery (HDR) provides integrated local backup and cloud disaster recovery for enterprise applications in a data center. The service protects mission-critical workloads in on-premises data centers and on Alibaba Cloud. It offers a recovery point objective (RPO) as low as seconds and a recovery time objective (RTO) in minutes. This ensures both data security and business continuity.
Prerequisites
Before you begin, you should understand the following basic concept: What is a data center?.
Core problems that HDR solves
Hybrid Disaster Recovery addresses the following core problems:
Application-level disaster recovery for business continuity: If a data center fails or requires extended maintenance, you can quickly recover applications in the cloud. This shortens business downtime and significantly reduces losses.
Data-level disaster recovery: You can back up your databases, virtual machines, and entire physical servers in your data center. The backup data is stored locally and automatically uploaded to the cloud. This protects your data if a major disaster occurs in your self-managed data center. It also provides efficient recovery from both local and cloud backups.
Supported data replication technologies
Continuous Data Replication (CDR): Provides an advanced disaster recovery plan for mission-critical enterprise applications. It offers an RPO and RTO in minutes. For more information, see Continuous Data Replication (CDR).
Cloud disk async replication: This method uses the async replication technology of Enhanced SSDs (ESSDs) to provide disaster recovery for complex ECS applications. It features an RPO as low as 15 minutes and an RTO in minutes. For more information, see Cloud disk async replication disaster recovery.
How to choose a replication technology?
In most cases, choose cloud disk async replication. You can choose CDR if you have a small number of ECS instances, a small total data volume, and a compatible operating system. The following table compares the two replication technologies.
Item | Cloud disk async replication (EBS Async) | Continuous Data Replication (CDR) |
Scenarios | Large number of ECS instances, diverse operating systems, and large data volumes. | 10 or fewer ECS instances. Small-scale applications, small ECS data volumes, and ECS disaster recovery with a CDR-compatible operating system. |
RPO, RTO | 15 minutes | 1 to 5 minutes (depends on the ECS write volume) |
Cost |
|
|
Intrusiveness | Non-intrusive | Requires a client to be installed, which consumes ECS resources. The client for Windows requires a restart after installation. |
Maximum replication throughput for a single disk | 100 MB/s | 30 MB/s |
Maximum replication throughput for a single site pair | No limit | 50 MB/s |
Sandbox drill | Not currently supported | Supported |
Operating system compatibility | Most operating systems | Limited to specific Windows and Linux systems. For more information, see Operating systems. |
Cloud disk type compatibility | ESSD | All types |
Terms
Before you use HDR, you should understand the following terms.
Concepts | Description |
Failover | Also known as disaster recovery. This is the process of recovering an application on Alibaba Cloud when the application in your data center fails. |
Failback | The process of migrating application data back to your own data center to resume application operations after the environment in your data center is restored. |
RPO | Recovery Point Objective (RPO) is the expected amount of data loss when an application fails. For example, an RPO of 15 minutes means that if an application fails, the data from the last 15 minutes cannot be recovered in the cloud. |
RTO | Recovery Time Objective (RTO) is the time required to recover an application in the cloud after a failure occurs. |
HDR pricing
HDR supports pay-as-you-go and subscription resource plans. For more information about the billable items and billing methods of HDR, see Billable items and billing methods.