Data Disaster Recovery
Baidu AI Cloud Object Storage (BOS) provides multi-AZ storage (intra-city redundancy) as well as cross-region and intra-region replication features, enabling disaster recovery at both intra-region and cross-region levels.
Multi-AZ storage
By enabling multi-AZ storage, BOS spreads user data across several availability zones within the same region. Even if one availability zone becomes inaccessible, normal data access remains uninterrupted. BOS multi-AZ storage provides 99.9999999999% (12 nines) data durability and 99.99% service availability.
BOS multi-AZ storage delivers robust disaster recovery capabilities for data centers. In situations like network disruptions, power outages, or disasters rendering a data center inoperative, it ensures continued strong-consistency services. The entire failover process is seamless to users, with no business interruptions or data loss, meeting the critical business system requirements for “Recovery Point Objective (RPO)” and “Recovery Time Objective (RTO)” of zero.
At present, BOS multi-AZ storage supports standard storage and infrequent access storage. A detailed comparison of metrics between these two storage classes is provided below:
| Comparison indicators | Standard storage-multi-AZ | Standard storage | Infrequent access storage-multi-AZ | Infrequent access storage |
|---|---|---|---|---|
| Availability | 99.995% | 99.99% | 99.995% | 99.95% |
| Minimum storage duration | None | None | 30 days | 30 days |
| Data restoration fees | None | None | Pay-as-you-go (GB) | Pay-as-you-go (GB) |
| Data access feature | Provides real-time access with millisecond-level latency. | Provides real-time access with millisecond-level latency. | Provides real-time access with millisecond-level latency. | Provides real-time access with millisecond-level latency. |
| Applicable scenarios | Ideal for mobile apps, video streaming on websites, data analysis, and other scenarios that demand extremely high reliability. | Suitable for mobile apps, video websites, data analysis, etc. | Fits scenarios like network disk storage, monitoring data, media asset backups, and other use cases requiring extremely high reliability. | Useful for network disk storage, monitoring data, media asset backups, and similar scenarios. |
| Scenario characteristics | Best suited for use cases with frequent data reads. | Best suited for use cases with frequent data reads. | Designed for scenarios where the average monthly access frequency is less than once. | Designed for scenarios where the average monthly access frequency is less than once. |
For function details, please refer to Multi-AZ Storage in the BOS Developer Guide.
Data synchronization
Data replication is an automatic and asynchronous process of copying files (objects) between storage spaces (buckets) within a BOS data center (region). It replicates object creation, updates, and deletions from the source storage space to the target storage space.
The data synchronization feature effectively supports intra-region, cross-region, cross-account, and other bucket replication needs. The objects in the target bucket are identical copies of the source bucket's objects, maintaining the same object name, metadata, and content, including creation time, owner, custom metadata, object ACL, and object content.
Application scenarios
- Efficient access with reduced latency: Cross-region access to buckets or objects may experience reduced speed due to geographical variation. The data synchronization function allows pre-synchronization of required data to the target region, to improve access efficiency.
- Cross-region disaster recovery: To meet compliance requirements in finance and government, etc., multiple data replicas are maintained within the same region. However, to prevent data center-level disasters, such as floods and earthquakes, it is also required to maintain a replica of data across regions. The data synchronization function of BOS provides this capability.
- Cross-account replication: It is suitable for multi-account scenarios of enterprises. For data security and backup considerations, enterprises may want to periodically sync data from Account A to Account B to replicate and share cross-account data.
- Cross-region data reuse: Due to business requirements, such as computing clusters built across regions needing to use the same set of data, the BOS data synchronization function can be used to replicate data copies.
- Data replication: Due to business requirements, data needs to be replicated from one BOS data center to another.
For function details, please refer to Data Synchronization in the BOS Developer Guide.
