Multi-Region Replication for Amazon Keyspaces (for Apache Cassandra)
You can use Amazon Keyspaces Multi-Region Replication to replicate your data with automated, fully managed, active-active replication across the Amazon Web Services Regions of your choice. With active-active replication, each Region is able to perform reads and writes in isolation. You can improve both availability and resiliency from Regional degradation, while also benefiting from low-latency local reads and writes for global applications.
With Multi-Region Replication, Amazon Keyspaces asynchronously replicates data between Regions, and data is typically propagated across Regions within a second. Also, with Multi-Region Replication, you no longer have the difficult work of resolving conflicts and correcting data divergence issues, so you can focus on your application.
By default, Amazon Keyspaces replicates data across three Availability
Zones
Topics
Benefits of using Multi-Region Replication
Multi-Region Replication provides the following benefits.
-
Global reads and writes with single-digit millisecond latency – In Amazon Keyspaces, replication is active-active. You can serve both reads and writes locally from the Regions closest to your customers with single-digit millisecond latency at any scale. You can use Amazon Keyspaces multi-Region tables for global applications that need a fast response time anywhere in the world.
-
Improved business continuity and protection from single-Region degradation – With Multi-Region Replication, you can recover from degradation in a single Amazon Web Services Region by redirecting your application to a different Region in your multi-Region keyspace. Because Amazon Keyspaces offers active-active replication, there is no impact to your reads and writes.
Amazon Keyspaces keeps track of any writes that have been performed on your multi-Region keyspace but haven't been propagated to all replica Regions. After the Region comes back online, Amazon Keyspaces automatically syncs any missing changes so that you can recover without any application impact.
-
High-speed replication across Regions – Multi-Region Replication uses fast, storage-based physical replication of data across Regions, with a replication lag that is typically less than 1 second.
Replication in Amazon Keyspaces has little to no impact on your database queries because it doesn’t share compute resources with your application. This means that you can address high-write throughput use cases or use cases with sudden spikes or bursts in throughput without any application impact.
-
Consistency and conflict resolution – Any changes made to data in any Region are replicated to the other Regions in a multi-Region keyspace. If applications update the same data in different Regions at the same time, conflicts can arise.
To help provide eventual consistency, Amazon Keyspaces uses cell-level timestamps and a last writer wins reconciliation between concurrent updates. Conflict resolution is fully managed and happens in the background without any application impact.
For more information about supported configurations and features, see Amazon Keyspaces Multi-Region Replication usage notes.
Capacity modes and pricing
For a multi-Region keyspace, you can either use on-demand capacity mode or provisioned capacity mode. For more information, see Read/write capacity modes in Amazon Keyspaces.
For on-demand mode, you're billed 1.25 write request units (WRUs) to write up to 1 KB of data per row. You're billed for writes in each Region of your multi-Region keyspace. For example, writing a row of 3 KB of data in a multi-Region keyspace with two Regions requires 7.5 WRUs: 3 * 1.25 * 2 = 7.5 WRUs. Additionally, writes that include both static and non-static data require additional write operations.
For provisioned mode, you're billed 1.25 write capacity units (WCUs) to write up to 1 KB of data per row. You're billed for writes in each Region of your multi-Region keyspace. For example, writing a row of 3 KB of data per second in a multi-Region keyspace with two Regions requires 7.5 WCUs: 3 * 1.25 * 2 = 7.5 WCUs. Additionally, writes that include both static and non-static data require additional write operations.
For more information about pricing, see Amazon Keyspaces (for Apache Cassandra) pricing