#aws #rds #aurora

Aurora is a proprietary technology from AWS for optimization and performance, better than RDS. Supports for: PostgreSQL and MySQL Storage is scaling same like RDS.

Replica

  • 15 replicas at a moment, replication process is faster.
  • Failover is instantaneous, it is high availability native.
  • Support for cross region replication

Specifications

  • Self healing
  • Auto expanding
  • Replication
  • Shared storage

Aurora Cluster, how it works

  • One master endpoints, will failover to another instance
  • One reader endpoints connects to one of multiple replica (via load balancing)

Features

  • Automatic failover
  • Backup and restore.
  • Isolation and security
  • Auto patching with zero downtime.
  • Restore at any point of time without using backup

Aurora Custom Endpoints

  • Define a subset of Aurora Instances as Custom Endpoint for some specific purpose like heavy analytic requires more powerful machine.
  • Reader endpoint usually is ignored after defining custom one.

Aurora Serverless

  • Automated database, instantiation, automated scaling based on actual usage.
  • Good for infrequent, unpredict workload
  • Use case: Use database only for short time.
  • Pricing based on seconds, so it can be cost-effective.

Aurora Multi Master

  • It’s only feature for Aurora.
  • Use case: Immediately failover for write node (HA). Normally, it takes short time to failover if we do not use this feature.
  • Every node has W/R and can be promoted to new master.

Global Aurora

  • Aurora cross region replica: For simple disaster recovery
  • Aurora Global Database: 1 primary region, up to 5 secondary (16 read replica per region), because of multi region so, replica is more faster, decreasing latency.

Aurora ML

  • Aurora integrates with ML services (SageMaker, Comprehend)
  • Just provide some SQL query and those service will returns some result for us based on its predictions
  • Use cases: ads targeting, fraud detection, product recommendations, sentiment analytic