All Products
Search
Document Center

ApsaraDB RDS:DuckDB-based analytical instances

Last Updated:Jun 04, 2026

RDS for MySQL DuckDB-based analytical instances use columnar storage and vectorized execution to accelerate complex analytical queries by 100x while significantly reducing storage costs.

Features

Two types are available:

  • DuckDB-based analytical primary instance

    • A standalone database instance similar to a regular RDS for MySQL primary instance. Available in Cluster Edition with multiple readable standby nodes.

    • Core features:

      • Retains all features of a regular RDS for MySQL instance and is compatible with existing tools and ecosystems. Overview of a DuckDB-based analytical primary instance.

      • Integrates the DuckDB analytical engine into the MySQL kernel, combining columnar storage and vectorized execution to deliver both strong transaction support and high-performance analytical processing.

  • DuckDB-based analytical read-only instance

    • A read-only instance attached to a regular RDS for MySQL High-availability Edition instance.

    • Core features:

      • Seamless data flow: Uses native binlog replication to automatically synchronize data and transform table schemas, eliminating the need for a separate synchronization link.

      • Integrated HTAP: Supports analytical queries through direct connections to the DuckDB read-only instance or automatic routing via a database proxy. This provides hybrid transactional and analytical processing (HTAP), ensuring TP performance of the primary instance while improving analytics efficiency.

Comparison item

DuckDB-based analytical primary instance

DuckDB-based analytical read-only instance

RDS for MySQL read-only instance

OLAP database

Scenario

Complex analytic queries

Complex analytic queries

Transaction processing

Complex analytic queries

Analytic query performance

Strong

Strong

Weak

Strong

Data synchronization method

DTS data synchronization link

Native replication based on binary logging

Native replication based on binary logging

DTS data synchronization link

MySQL compatibility

Data type

Fully compatible

Fully compatible

Fully compatible

Incompatible (requires field mapping)

SQL syntax

Highly compatible (> 99.9%)

Highly compatible (>99.9%)

Fully compatible

Incompatible (requires SQL rewrite)

DDL

Highly compatible

Highly compatible (supports automatic rebuilding)

Fully compatible

Partially compatible

O&M costs

Low

Low

Low

High

Deployment model

Standalone deployment

Attached to a regular RDS for MySQL instance

Attached to a regular RDS for MySQL instance

Standalone deployment

Supported edition

Cluster Edition

High-availability Edition

Basic Edition, High-availability Edition

-

Scenarios

  • Aggregate analysis: Efficient aggregate queries for real-time analysis, such as log data analytics.

  • Multi-table join queries: Significantly improves MySQL analytical performance for workloads involving multi-table JOIN operations.

Technical principles

What is DuckDB?

DuckDB is a standalone, embedded OLAP database that balances high-performance analytics with transaction processing:

  • High-performance analytics: Columnar storage accelerates aggregate queries, and the vectorized execution engine processes data in batches.

  • Strong transaction support: Full ACID compliance with MVCC for efficient concurrent reads and writes.

Technical optimizations

ApsaraDB RDS integrates the DuckDB engine into the MySQL ecosystem, combining high-performance analytics with enterprise-grade data reliability and consistency.

Shared optimizations

  • Storage engine optimization

    • Encapsulates DuckDB columnar storage as a MySQL transactional storage engine, using the standard data dictionary for metadata management. Fully compatible with primary instance transaction semantics.

    • Enhances binlog-based synchronization and uses DuckDB write-ahead logging (WAL) for transaction persistence, ensuring data consistency between the RDS instance and the DuckDB analytical instance.

  • Compute engine enhancement

    • Integrates the DuckDB optimizer, vectorized execution engine, and JIT compiler to improve complex query performance by 100x compared with InnoDB.

    • Supports 99.9% of MySQL syntax and functions, so existing queries run without modification.

    • The Result Translator converts DuckDB results into MySQL protocol format for seamless client integration without application changes.

  • Query acceleration

    • Columnar storage: Data stored in DuckDB's native columnar format improves aggregate query performance by over 100x. For example, a SUM query runs 100x faster than on InnoDB.

    • Automatic hot data caching: Uses DuckDB's buffer pool to cache frequently queried data, providing stable performance in high-concurrency scenarios.

Read-only instance optimizations

  • Data synchronization performance improvement

    • Batches high-frequency small transactions into larger ones to reduce write latency.

    • Idempotent replay: Performs idempotence checks on binlog events to ensure strong data consistency.

  • DDL synchronization optimization

    • Automatically routes DuckDB-native DDL operations (such as creating tables and adding or removing columns) directly to the DuckDB engine without conversion.

    • For unsupported DDL operations, automatically triggers a table rebuild to prevent replication interruptions.

  • Resource isolation: The DuckDB read-only instance runs in isolation from the primary instance, so analytical queries do not affect online transaction processing.

Query workflow

  1. SQL parsing: A user's SQL query is parsed by the standard MySQL parser.

  2. SQL execution: The parsed query is routed to the DuckDB compute engine for execution.

  3. Data retrieval: Columnar data is retrieved from the DuckDB storage engine.

  4. Format conversion: The computation result is converted by the Result Translator component into the MySQL protocol format and then sent to the client.

Get started