Overview of Amazon DocumentDB Global Clusters - Amazon DocumentDB

Overview of Amazon DocumentDB Global Clusters

What is a global cluster?

A global cluster consists of one primary region and up to five read-only secondary regions. You issue write operations directly to the primary cluster in the primary region and Amazon DocumentDB automatically replicates the data to the secondary regions using dedicated infrastructure. Latency is typically under a second.

How are global clusters useful?

  • Recovery from region-wide outages — The secondary clusters allow you to quickly recreate a global cluster with a new primary cluster, without data loss.

  • Global reads with local latency — If you have offices around the world, you can use a global cluster to keep your main sources of information updated in the primary region. Offices in your other regions can access the information in their own region, with local latency.

  • Scalable secondary clusters — You can scale your secondary clusters by adding more read-only instances to a secondary region. The secondary cluster is read-only, so it can support up to 16 read-only replica instances rather than the usual limit of 15 for a single cluster.

  • Fast replication from primary to secondary clusters — The replication performed by a global cluster has little performance impact on the primary database cluster. The resources of the DB instances are fully devoted to serve application read and write workloads.

What are the current limitations of global clusters?

  • Global clusters are not supported on Amazon DocumentDB v3.6.

  • Global clusters are not supported on t3 and r4 instance types.

  • In the event of a regional failover, you must manually promote a secondary cluster to become the primary cluster, and modify your application to point to the new primary cluster.

  • Only the primary cluster performs write operations. Clients that perform write operations connect to the cluster endpoint of the primary cluster.

  • You can have a maximum of five secondary regions and one primary region for your cluster.

  • A secondary cluster cannot be stopped. A primary cluster cannot be stopped if it has secondary clusters associated with it. Only a regional cluster that has no secondary clusters can be stopped.

  • Replicas attached to the secondary cluster can restart under certain circumstances. If the primary region's instance restarts or fails over, replicas in the secondary region also restart. The cluster is then unavailable until all replicas are back in sync with the primary database cluster's writer instance. This behavior is expected. Be sure that you understand the impact to your global cluster before making changes to your primary cluster.

  • You cannot use AWS CloudFormation to manage global clusters.

  • You cannot use change streams on secondary clusters.