Amazon Aurora

Amazon Aurora is a MySQL- and PostgreSQL-compatible relational database engine built for the cloud. It combines the performance and availability of commercial databases with the simplicity and cost-effectiveness of open-source engines. Aurora delivers up to five times the throughput of MySQL and three times the throughput of PostgreSQL — without requiring changes to your application code.

---

Architecture — What Makes Aurora Different

Shared Distributed Storage

Traditional RDS instances store data on a single EBS volume attached to the primary instance. Aurora decouples compute from storage:

• Data is stored in a shared, distributed storage layer spread across three Availability Zones
• Six copies of your data are maintained — two copies in each AZ
• The storage volume grows automatically in 10 GiB increments up to 128 TiB
• You never need to provision or manage storage; you pay only for what you use

How the Six Copies Protect You

Scenario	Copies Lost	Service Impact
One AZ fails	2	No impact — Aurora can still read and write
Two copies fail (any AZ)	2	No impact — quorum maintained
Three copies fail	3	Reads still work, writes paused until recovery

Aurora uses a quorum-based protocol: writes require 4 of 6 copies, reads require 3 of 6 copies. This allows Aurora to tolerate failures without data loss.

Continuous Backup to S3

Aurora continuously backs up data to Amazon S3 in real time. There is no performance impact and no backup window. You get point-in-time recovery (PITR) to any second within your retention period (up to 35 days).

---

Aurora Cluster Topology

An Aurora cluster consists of:

1. One primary (writer) instance — handles reads and writes
2. Up to 15 Aurora Replicas — handle read traffic with typically < 10 ms replica lag
3. Cluster endpoint — DNS name that always points to the primary
4. Reader endpoint — DNS name that load-balances connections across all replicas
5. Custom endpoints — target specific subsets of instances (e.g., route analytics queries to larger instances)

             ┌─────────────────────────────────────────┐
             │          Aurora Storage Layer            │
             │   (6 copies across 3 AZs, auto-grows)   │
             └────────────┬────────────┬───────────────┘
                          │            │
              ┌───────────┴──┐   ┌─────┴──────────┐
              │   Primary    │   │  Replica × 15   │
              │  (writer)    │   │  (readers)       │
              └──────────────┘   └─────────────────┘
                     ▲                   ▲
             Cluster endpoint      Reader endpoint

---

Aurora vs Standard RDS — Feature Comparison

Feature	Standard RDS	Aurora
Storage replication	Single AZ (EBS)	6 copies across 3 AZs
Auto-growing storage	Manual or autoscale	Automatic (10 GiB increments)
Max storage	64 TiB	128 TiB
Replicas	Up to 5	Up to 15
Replica lag	Seconds	< 10 ms
Failover time	60–120 s	~30 s
Backups	Snapshot + transaction logs	Continuous to S3
Backtrack	Not available	Yes (MySQL-compatible)
Serverless option	No	Yes (Aurora Serverless v2)

---

Aurora Serverless v2