CodeParkAurora works like MySQL or PostgreSQL, but it stores data in a different way. Your driver and ORM work without changes.
In standard RDS, each instance has its own disk. In Aurora, all instances in a cluster share one distributed storage volume. This volume keeps six copies of the data across three AZs (Availability Zones: separate data centres). It grows automatically.
Replicas read the same shared storage, so they do not need to replay a full copy of the data. Because of this:
- Replica lag is usually low.
- Failover to a replica is fast.
A cluster has two endpoints. The writer endpoint sends writes to the primary. The reader endpoint spreads connections across the replicas. You can have up to 15 Aurora replicas.
Aurora also offers:
- Serverless v2, which scales capacity up and down in small steps.
- Global Database, for reads in other Regions and DR (disaster recovery).
Aurora usually costs more than plain RDS. So it is worth it for large workloads or ones that must stay available.
- Aurora is compatible with MySQL and PostgreSQL.
- All instances share distributed storage with six copies across three AZs, and it grows automatically.
- Replicas share the same storage, so lag is low and failover is fast.
- A cluster has a writer endpoint and a reader endpoint, and up to 15 replicas.
- Aurora offers Serverless v2 and Global Database, and costs more than plain RDS.