The DuckDB-based analytical primary instance runs complex analytical queries alongside transactional workloads in a single MySQL-compatible database. It embeds the DuckDB engine directly in the MySQL kernel, using columnar storage and vectorized execution to run analytical queries over 100x faster than the InnoDB engine—without a separate analytics cluster or ETL pipeline.
Use cases
Real-time reporting on live data: Run dashboards, aggregations, and ad hoc queries directly on live transactional data without affecting your OLTP workload.
Replacing a standalone analytics database: Eliminate a separate analytics system. The instance handles both workloads, and data written by transactions replicates to the columnar store via MySQL binlog, keeping analytical results current.
Cost-sensitive analytical workloads: Columnar storage compresses data by over 50% compared to InnoDB, reducing storage costs while maintaining query performance.
How it works
The DuckDB engine is deeply integrated within the MySQL kernel. Both the DuckDB columnar engine and the InnoDB engine run within the same MySQL kernel, so you connect and query the instance exactly as you would a standard RDS for MySQL instance.
Data replication: Transactions replicate to the columnar store via MySQL binlog. The instance supports transactions at the Snapshot Isolation level, ensuring data reliability and consistency.
For the full technical architecture, see Technical principles of DuckDB analytical instances.
Core advantages
100x faster analytics: The DuckDB engine uses columnar storage, vectorized execution, parallel processing, and efficient memory management to accelerate complex queries. Standard MySQL queries run unaffected on InnoDB.
No ETL required: Transactions replicate to the columnar store via MySQL binlog. Analytical queries always read current data.
Over 50% storage reduction: Columnar storage achieves a high data compression ratio, reducing disk space compared to InnoDB.
MySQL ecosystem compatibility: Fully compatible with the MySQL protocol, syntax, and data types. Supports transactions at the Snapshot Isolation level and uses MySQL binlog for replication.
High availability with crash safety: Cluster Edition uses a one-primary, multiple-secondary architecture with automatic failover. All secondary nodes are readable. The crash-safe kernel preserves data durability in failure scenarios, and point-in-time restore is supported.
Simple architecture: Uses the same database kernel as RDS for MySQL with no external dependencies.
Deployment model
| Attribute | Value |
|---|---|
| Major engine version | MySQL 8.0 |
| Billing | pay-as-you-go and subscription |
| Edition | Cluster Edition only. One primary and multiple secondary nodes; all secondary nodes are readable. Add or remove nodes on demand. |
| Storage type | Premium ESSD |
| Storage capacity | 20 GB to 128,000 GB |
Features
"Supported" means the feature is available. "Not supported" means it is unavailable.