xCluster - Native asynchronous replication

By default, YugabyteDB provides synchronous replication and strong consistency across geo-distributed data centers. A cross-cluster (xCluster) deployment provides high throughput asynchronous replication across two data centers or cloud regions.

xCluster provides the following benefits:

• Low write latency: With synchronous replication, each write must reach a consensus across a majority of data centers. This can add tens or even hundreds of milliseconds of extra latency for writes in a multi-region deployment. Asynchronous replication with xCluster reduces this latency by eliminating the need for immediate consensus across regions.
• Only two data centers needed: With synchronous replication, to tolerate the failure of f fault domains, you need at least 2f + 1 fault domains. Therefore, to survive the loss of one data center, a minimum of three data centers is required, which can increase operational costs. For more details, see fault tolerance. However, with async replication you can achieve multi-region deployments with only two data centers.
• Disaster recovery: By replicating data asynchronously to a secondary region, xCluster provides a robust disaster recovery solution, allowing for quick failover and minimal data loss in case of a regional outage.

xCluster is always active-active, allowing you to read data from any region. It supports both single-master and multi-master deployment options, offering flexibility to balance write performance and consistency guarantees.

Active-active single-master xCluster:

Active-active multi-master xCluster:

For more information, see the following:

• xCluster architecture
• xCluster vs Geo-Partitioning vs Read Replicas
• Deploy xCluster
• Change data capture (CDC)