Overview
AWS proprietary database — not open source, but compatible with MySQL and PostgreSQL drivers.
| Feature | Detail |
|---|---|
| Compatibility | MySQL and PostgreSQL drivers |
| Performance | 5x MySQL on RDS, 3x PostgreSQL on RDS |
| Storage | Auto-grows from 10 GB to 128 TB (256 TiB on newer versions) |
| Read Replicas | Up to 15 (sub 10ms replica lag) |
| Failover | Instantaneous (< 30 seconds avg) |
| High Availability | Native (cloud-native by default) |
| Cost | ~20% more than RDS, but more efficient at scale |
High Availability & Read Scaling
Aurora stores 6 copies of your data across 3 AZs:
| Operation | Copies Needed | Implication |
|---|---|---|
| Writes | 4 out of 6 | Survives losing 1 AZ |
| Reads | 3 out of 6 | Highly available for reads |
AZ 1 AZ 2 AZ 3
┌────────┐ ┌────────┐ ┌────────┐
│ Copy 1 │ │ Copy 3 │ │ Copy 5 │
│ Copy 2 │ │ Copy 4 │ │ Copy 6 │
└────────┘ └────────┘ └────────┘
└──────────┼──────────┘
Shared Storage Volume
(auto expanding, self-healing,
peer-to-peer replication)
- Self-healing: corrupted data repaired via peer-to-peer replication
- Storage uses hundreds of volumes (not just one)
- Only one master takes writes; up to 15 read replicas serve reads
- Any read replica can become master on failover
- Supports cross-region replication
6 copies across 3 AZs
Aurora Cluster Architecture
┌──────────────┐
│ Client │
└──┬───────┬───┘
│ │
Writer │ │ Reader
Endpoint │ │ Endpoint
│ │
┌──────▼──┐ ┌──▼──────────────┐
│ Master │ │ Read Replicas │
│ (write) │ │ (1-15, auto │
│ │ │ scaling) │
└────┬────┘ └────────┬─────────┘
│ │
┌──────▼───────────────▼──────┐
│ Shared Storage Volume │
│ 10 GB → 128 TB auto (256 TiB on newer versions) │
└─────────────────────────────┘
| Endpoint | Purpose |
|---|---|
| Writer Endpoint | DNS name always pointing to the master (survives failover) |
| Reader Endpoint | Connection load balancing across all read replicas |
Reader endpoint load balancing happens at the connection level, not the statement level.
Aurora cluster endpoints
Key Features Summary
| Feature | Detail |
|---|---|
| Automatic failover | < 30 seconds |
| Backup & recovery | Built-in |
| Push-button scaling | Auto scaling replicas |
| Automated patching | Zero downtime |
| Backtrack | Restore to any point in time (not backup-based) |
| Advanced monitoring | Built-in |
| Cross-region replication | Supported |
Unlike regular restore, backtrack lets you go back and forth in time. "Go to yesterday 4 PM... actually, 5 PM" — both work without creating a new DB.
Questions & Answers
Q1: Is Aurora open source?
Answer:
No. Aurora is AWS proprietary, but it's compatible with MySQL and PostgreSQL drivers. Your existing MySQL/PostgreSQL applications can connect to Aurora without changes.
Q2: How much faster is Aurora compared to RDS?
Answer:
5x faster than MySQL on RDS and 3x faster than PostgreSQL on RDS, due to cloud-optimized architecture.
Q3: How does Aurora storage work?
Answer:
Auto-grows from 10 GB to 128 TB (256 TiB on newer Aurora versions like PostgreSQL 16.9+). Stores 6 copies across 3 AZs. Uses hundreds of volumes with self-healing and peer-to-peer replication.
Q4: How many copies does Aurora need for writes and reads?
Answer:
- Writes: 4 out of 6 copies needed
- Reads: 3 out of 6 copies needed This means Aurora survives losing an entire AZ.
Q5: What is the Aurora Writer Endpoint?
Answer:
A DNS name that always points to the current master instance. Even if the master fails over, the writer endpoint automatically redirects to the new master.
Q6: What is the Aurora Reader Endpoint?
Answer:
A DNS name that provides connection load balancing across all read replicas. Load balancing happens at the connection level, not statement level.
Q7: How fast is Aurora failover?
Answer:
Instantaneous — less than 30 seconds on average. Much faster than Multi-AZ failover on regular RDS.
Q8: How many read replicas can Aurora have?
Answer:
Up to 15 read replicas with sub 10ms replica lag. They can also have auto scaling.
Q9: What is Aurora Backtrack?
Answer:
A feature that lets you restore data to any point in time without relying on backups. You can go back and forth in time (e.g., "yesterday 4 PM" then change to "5 PM").
Q10: How much more does Aurora cost compared to RDS?
Answer:
About 20% more than RDS, but it's much more efficient at scale due to better performance, auto-scaling storage, and cloud-native architecture.
Check yourself
Quiz
Flashcards
Click the card (or focus it), then use ← → to move and Space to flip.