Building resilient and disaster-tolerant Amazon Neptune deployments
The Amazon global infrastructure is built around Amazon Regions and Availability Zones. Amazon Regions provide multiple physically separated and isolated Availability Zones, which are connected with low-latency, high-throughput, and highly redundant networking. With Availability Zones, you can design and operate applications and databases that automatically fail over between Availability Zones without interruption. Availability Zones are more highly available, fault tolerant, and scalable than traditional single or multiple data center infrastructures.
An Amazon Neptune DB cluster can only be created in an Amazon VPC that has at least two subnets in at least two Availability Zones. By distributing your cluster instances across at least two Availability Zones, Neptune helps ensure that there are instances available in your DB cluster in the unlikely event of an Availability Zone failure. The cluster volume for your Neptune DB cluster always spans three Availability Zones to provide durable storage with less possibility of data loss.
For more information about Amazon Regions and Availability Zones, see Amazon Global Infrastructure