A decision framework for Amazon Neptune availability

This article provides a decision framework for choosing Amazon Neptune availability patterns based on RTO, RPO, and operational complexity requirements.

• Neptune Multi-AZ baseline provides 99.99% availability with automatic failover under 120 seconds
• Global Database enables multi-Region deployments with replication lag typically under one second
• Manual cross-Region failover (switchover/failover) prevents split-brain scenarios and false positives
• Write durability patterns: SQS queues accept delayed visibility; DynamoDB Global Tables provide near-zero RPO
• Single-Region patterns include blue/green deployments, ElastiCache caching, and graceful feature degradation
• Operational best practices: maintain two readers across AZs, monitor GlobalDbProgressLag, conduct quarterly Game Day tests
• Layer patterns progressively from simplest (Multi-AZ) to most complex (Global Database + DynamoDB WAL)

Organizations should identify RTO/RPO targets, evaluate current deployment against Multi-AZ baseline, then layer additional patterns only as requirements demand.