Overview
Cross-cloud replication is an advanced feature provided by COS that enables object movement operations between different cloud environments. By configuring cross-cloud replication rules, users can automatically and asynchronously copy incremental objects from the source bucket to the destination bucket.
Once cross-cloud replication is enabled, COS precisely copies the object content in the source bucket, including the object metadata and version ID, ensuring that the copied object replicas have completely consistent cloud data with the source objects. In addition, overwrite operations on objects in the source bucket will be synchronized to the destination bucket. However, it should be noted that deletion operations in the source bucket will not be synchronized through cross-cloud replication.
Note:
- Cross-cloud replication supports any COS that supports the S3 protocol as the destination.
- Cross-cloud replication has basically no requirements for the destination bucket. For example, the destination bucket can be a bucket under the same account, a bucket in a different region, or a bucket in a different cloud.
- Cross-cloud replication supports historical data synchronization, but currently requires manual configuration by the Ops personnel.
Applicable Scenarios
- Implementing cross-region disaster recovery: COS provides 11 nines of durability for object data, but there is still a slight chance of data loss due to force majeure such as wars and natural disasters. If you want to avoid data loss and explicitly have a separate replica of data in a different region or cloud, you can use cross-cloud replication to implement cross-region disaster recovery. In this way, when an IDC is damaged due to force majeure, the IDC in another region or cloud can still provide data replicas for your use.
- Meeting compliance requirements: COS provides methods such as multiple replicas and erasure coding in physical disks for data by default to ensure data availability. However, some industries may have compliance requirements stipulating that you preserve data replicas across different storage regions or clouds. Therefore, enabling cross-cloud replication allows you to copy data across different storage regions or clouds to meet such requirements.
- Reducing access latency: When your end users access your objects from different geographic locations, you can use cross-cloud replication to maintain the object replicas in the available storage region or cloud closest to your end users' geographic locations. This will minimize your end users' access latency and improve your product experience.
- Meeting special technical requirements: When you have compute clusters in two regions or clouds, and these clusters need to process the same set of data, you can use cross-cloud replication to maintain object replicas in both regions.
- Migrating and backing up data: You can copy your business data from one available region or cloud to another according to the development needs of your business to achieve data migration and backup.
Must-Knows
Replication Time Limit
The time it takes for COS to copy an object depends on factors such as the object size, the distance between the storage regions, and the upload method of the object. Based on the above factors, the synchronization time varies from a few minutes to a few hours.
- Object size: It takes more time to copy larger objects. For them, it is recommended that you use multipart upload to reduce their upload and synchronization time.
- Distance between the storage regions or clouds: It takes more data transfer time for a longer distance between the regions or clouds.
- Upload method of the object: A simple upload does not support concurrent uploads and can only upload or download data serially over a single connection, but a multipart upload supports concurrent uploads. Therefore, using multipart upload for large files will speed up upload and cross-cloud replication.