xCluster Replication: Discover The Power and Flexibility of Asynchronous Replication
summary
YugabyteDB’s xCluster replicates data asynchronously between two clusters to enable low-latency local reads and act as DR copies, batching WAL changes per tablet and applying them into the storage layer. Demo shows bidirectional inserts updates and deletes replicate successfully; expected inter-coast lag is about 70–100 ms. Conflicts use last-writer-wins and triggers do not fire on the destination. Current limitations around transactional atomicity and global ordering are resolved in release 2.17, and easier failover and DDL sync are coming.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 0:48Normal Replication
- 5:54Use Cases
- 10:50How xCluster Works
- 14:09Asynchronous Replication Drawbacks
- 16:43xCluster 217
- 18:22Upcoming improvements
- 20:43Demo
related talks
01h 00mParallel Asynchronous Replication b/t YDB Database Instances
27mSparking our way to YugabyteDB: Our journey from ScyllaDB to
24mFailure is Not an Option: Highly Available Distributed SQL
47mAnatomy of a Distributed SQL Database - YugabyteDB
29mHow Helpshift Accelerated Growth by Migrating to
29mScalable Product Mappings
25mThe future of databases: Combining operational and analytical data at massive scale
21mHow Safe is Asynchronous Master-Master Setup?