Aurora¶

What is Aurora?¶

• Compatible with Postgres & MySQL.
• Proprietary AWS technology.
• Cloud optimized.
• 5x performance improvement over MySQL on RDS, and 3x Postgres on RDS.
• Uses auto-scaling storage.
• 10GB, upto 64TB.
• Supports up to 15 read rpelicates (sub 10ms replica lag).
• Instantaneous failover.
• HA by default.
• 20% more than RDS.
• Only one Aurora instance (master) takes writes.

High Availability¶

• Uses shared storage volume, to store 6 copies of data across 3 AZs.
• Only needs 4 copies out of 6 for writes, and 3 out of 6 for reads.
• Self healing with peer-to-peer replication to remediate bad data.
• Storage striped across 100s of volumes.
• Automated failover for master in less than 30seconds.
• Supports cross region replication.

Read Scaling¶

• Read replicas support auto-scaling.

Aurora DB Cluster¶

• DNS name (writer endpoint) always pointing to master to write data.
• Reader endpoint manages connection load balancing to connect client to one of the read replicas.
• Load balancing is done at the connection level.

Features¶

• Automatic failover.
• Backup & recovery.
• Isoltion & security.
• Industry compliance.
• Push button scaling.
• Automated patching with no downtime.
• Advanced monitoring.
• Routine maintenance.
• Backtrack.

Security¶

• Encryption at rest via KMS.
• Automated backups, snapshots and replicas are encrypted.
• Encryption in transit via SSL.
• Can authenticate using IAM tokens.
• User responsible for configuring security groups.
• Can't SSH into instance.

Aurora Serverless¶

• Automated db instantiation and auto-scaling based on actual usage.
• Good for infrequent, intermittent or unpredictable workloads.
• Removes the need for capacity planning.
• Pay per second, can be more cost effective.
• Clients connect to a proxy fleet, that proxy the comms to the Aurora instances.

Global Aurora¶

• Cross-region read replicas.
• Good for DR.
• Easy to setup.
• Aurora Global Database
• 1 primary region for read/write.
• Up to 5 secondary read only regions.
• Replication lag < 1second.
• Up to 16 read replicas per secondary region.
• Decreases latency.
• Promoting to another region due to a DR event takes < 1min.