Amazon Aurora Review
March 11, 2019
Amazon Aurora Review
Score 9 out of 10
Overall Satisfaction with Amazon Aurora
Amazon Aurora is a PaaS database product from AWS that is a drop-in replacement for existing workloads utilizing either a MySQL or PostgreSQL backend that improves upon the database engine performance of those open source projects. We leverage Aurora for its simple scaling without having to take a cluster down, and find its auto-scaling storage to be a better fit for our workloads than having to guess ahead of time and over-provision.
- Performance: We utilize Aurora as a PostgreSQL replacement, and Aurora's throughput is up to 3 times higher.
- Simple Instance Auto-Scaling: We can scale the underlying database engine up or down with no down time.
- Auto-Growing Storage: Rather than having to over-provision, Aurora automatically adds blocks of 10GB to your storage cluster up to multiple terabytes of storage.
- Support for additional engines: Right now, Aurora is limited to MySQL and PostgreSQL.
- PostgreSQL-specific Instance Types: The PostgreSQL has high minimum instance type variants; while MySQL can take advantage of t3 instances, the minimum PostgreSQL instance is too large for lower-budget workflows and tests/debugging.
- Has enabled us to not pay for over provisioned database storage that we may or may not need thanks to auto-scaling features.
- Per-second replica ability gives us peace of mind.
Aurora is a terrific drop-in replacement for many workloads that are now running on Amazon's RDS product. Aurora comes with significant performance improvements and additional features, namely the simple, no down time compute scaling, storage auto-scaling, extremely low latency read replicas, and incremental backups that allow to rollback to any second in the past within the storage window.
For workloads that already use, or plan on using, MySQL or PostgreSQL, Aurora is our new go-to favorite deployment option for projects on AWS. The best use cases for Aurora will be substantial workloads that are well-suited to the simple scaling controls (both from an instance type perspective, as well as storage perspective), and will benefit from Aurora's simple, very low latency read replicas. Aurora is extremely fault tolerant and has improved self-healing ability.