6.69 Amazon aurora overview

This lesson is a complete overview of Amazon Aurora. Aurora is AWS's proprietary database engine, compatible with MySQL and PostgreSQL drivers — your existing client code keeps working. Aurora is cloud-optimized and offers up to 5x MySQL performance on RDS and up to 3x PostgreSQL performance. Storage scales automatically by 10 GB increments up to 64 TB, and you can run up to 15 read replicas — vs 5 on RDS MySQL — with a replication lag below 10 ms.

Architecture highlights

  • Data is stored as 6 copies across 3 availability zones, with self-healing storage.
  • 4 of 6 writes are required to acknowledge a write; 3 of 6 reads are required to serve a read.
  • Master failover is automated in under 30 seconds.
  • A writer endpoint DNS name always points to the current master, even after failover.
  • A separate reader endpoint load-balances queries across read replicas (which themselves can auto-scale).
  • Native cross-region replication for disaster recovery.

Aurora costs about 20% more than RDS but its efficiency and faster recovery often justify the extra spend. The engine supports the same security model as RDS: at-rest encryption with KMS (AES-256), in-transit encryption with SSL, IAM-based authentication tokens, and security groups. Snapshots, replicas and backups inherit encryption from the cluster setting.

Aurora also offers a Serverless mode for sporadic workloads — capacity scales automatically per second based on demand, ideal for unpredictable usage patterns. For globally distributed apps, Aurora Global Database gives you one primary region and up to five secondary regions, with cross-region replication under one second and up to 16 read replicas per region. Disaster recovery RTO is about one minute. We will cover the creation of an Aurora cluster in the next hands-on lab.

Summary

Amazon Aurora is a proprietary AWS relational database service supporting PostgreSQL and MySQL, delivering up to 5x performance improvement over MySQL RDS and 3x over PostgreSQL RDS. It features automatic storage scaling from 10 GB to 64 TB, up to 15 read replicas, and built-in high availability through data replication across three availability zones with automatic failover in under 30 seconds. Aurora uses a shared storage architecture with automatic maintenance, self-healing capabilities, and dedicated read/write endpoints for seamless routing.

Key points

  • Performance optimized for cloud: up to 5x faster than MySQL RDS and 3x faster than PostgreSQL RDS
  • Automatic storage scaling up to 64 TB with self-healing disks and automatic maintenance
  • High availability: data replicated across 6 copies in 3 availability zones with automatic failover in under 30 seconds
  • Supports up to 15 read-only replicas distributed across zones with automatic read endpoint load balancing
  • Shared storage with sub-10-millisecond replication lag between availability zones
  • Approximately 20% higher cost than RDS but significantly more efficient for production workloads

FAQ

How does Aurora achieve high availability?

Aurora replicates data across 6 copies distributed in 3 availability zones. Write operations require 4 out of 6 copies, while reads require only 3, enabling automatic failover and reconstruction of failed instances in under 30 seconds.

How many read replicas can Aurora support?

Aurora supports up to 15 read-only replicas automatically distributed across availability zones. These are managed by Aurora's reader endpoint, which load-balances read queries across available replicas.

How does Aurora storage scaling work?

Aurora automatically scales shared storage from 10 GB up to 64 TB without requiring manual intervention. Storage is distributed across availability zones with automatic self-healing and peer-to-peer replication of failed volumes.