PROVIDER Β· AWS cloud-native relational database
Amazon Aurora
Amazon Aurora is AWS's cloud-native relational database compatible with MySQL and PostgreSQL, separating compute from storage that grows to 128 TiB with up to 15 read replicas, the default managed primary on AWS.
4.6pts
Pros
Cons
Product Information
Detailed Notes
Overview
Amazon Aurora became generally available in 2015 as AWS's cloud-native relational database compatible with MySQL and PostgreSQL protocols, separating compute from storage that scales automatically to 128 TiB with six copies across three availability zones.
For Backend & Database work its value is raising the ceiling of managed MySQL, since a log-is-the-database design ships log records instead of pages for several times the throughput of standard MySQL, and 15 read replicas, cross-region Global Database, and Serverless v2 second-level scaling cover workloads from websites to core transactions.
The cost is lock-in and pricing, since it runs only in AWS regions and cross-cloud migration means rebuilding, billing combines instance hours with I/O and storage so I/O-heavy workloads surprise budgets unless using I/O-Optimized storage, and Serverless v2 ACU minimums need careful modelling for intermittent load.
Core Strengths
- Storage auto-grows to 128 TiB without pre-partitioning or manual resizing
- Six copies across three AZs with replica failover usually inside 30 seconds
- Up to 15 read replicas far exceed standard RDS read scaling
- Global Database provides low-latency cross-region replication and disaster recovery
Product Ecosystem
MySQL and PostgreSQL compatible editions
Two branches align with MySQL 5.7 or 8.0 and PostgreSQL 13+ features respectively, with no direct migration across branches.
Aurora Serverless v2
Scales in ACUs by the second for spiky traffic and dev environments, with minimum capacity billing to watch.
Global Database and I/O-Optimized
The former adds cross-region DR and local reads, the latter bills on throughput and cuts cost for I/O-heavy workloads.
Drawbacks
- Runs only inside AWS, so cross-cloud or hybrid needs extra architecture
- Instance plus I/O and storage billing surprises I/O-heavy budgets
- Minor divergences from stock MySQL and PostgreSQL trip some third-party tools
- Egress is expensive and cross-region replication carries ongoing cost
Use Cases
- Websites and SaaS primaries on AWS (β β β β β ) - Managed with a compatible ecosystem
- Read-heavy workloads (β β β β β ) - Up to 15 replicas
- Multinational local reads (β β β β ) - Low-latency Global Database
- Cloud-neutral architectures (β β ) - Confined to AWS regions
Pricing
| Plan | Price | Scope |
|---|---|---|
| Serverless v2 | Per ACU-second from about USD 0.12 | Auto-scaling for variable load |
| Provisioned instances | Hourly per instance class | I/O and storage billed separately |
| I/O-Optimized storage | Per throughput and storage | Lower total cost for heavy I/O |
Aurora bills instance hours, I/O requests, storage, and backups with I/O the most under-estimated, so read-heavy workloads should offload to replicas and I/O-intensive ones usually save 20 to 40 percent on I/O-Optimized storage.
FAQ
-
Aurora or RDS for MySQL? Choose Aurora for many replicas, auto-growing storage, or cross-region DR, while small steady workloads cost less on RDS instances.
-
Does Serverless v2 always save money? Only under variable or idle load, since provisioned instances price lower for sustained demand, so model real traffic across minimum and maximum ACUs.
-
Can it move to MySQL on another cloud? Yes through logical replication or export and import given protocol compatibility, though performance assumptions and replica behaviour must be retested.