alibaba/AliSQL

AliSQL is a MySQL branch originated from Alibaba Group. Fetch document from Release Notes at bottom.

What it solves

AliSQL addresses the limitations of standard MySQL by adding native support for high-performance analytical queries and AI-driven vector search, while maintaining full compatibility with the MySQL client protocol.

How it works

It is a branch of MySQL 8.0.44 that integrates several specialized engines and optimizations:

  • DuckDB Integration: Incorporates a DuckDB storage engine to provide columnar execution and storage, significantly speeding up analytical aggregations compared to InnoDB.
  • Vector Search: Implements a VECTOR(N) data type and HNSW (Hierarchical Navigable Small World) indexes to enable efficient Approximate Nearest Neighbor (ANN) searches using cosine and Euclidean distance.
  • Data Management: Includes "Native Flashback" for querying historical data using timestamps and various binlog optimizations to reduce I/O amplification for large transactions.

Who it’s for

Database administrators and developers who need the familiarity and ecosystem of MySQL but require advanced analytical capabilities or the ability to store and store and search embeddings for AI applications.

Highlights

  • Analytical Speedup: Achieves over 200x speedup over InnoDB for specific TPC-H benchmarks via the DuckDB engine.
  • C-Compatible: Works with existing MySQL clients and drivers.
  • High-Dimensional Vectors: Supports vector indexes for embeddings up to 16,383 dimensions.
  • Historical Queries: Native support for AS OF TIMESTAMP queries to access previous data states.

Related

  • Project
    VexDB-THU/VexDB-Lite

    A high-performance vector similarity search engine for PostgreSQL, DuckDB, and SQLite that uses a custom graph index and quantization to enable efficient ANN search.

  • Project
    MariaDB/server

    An open-source relational database that provides MySQL compatibility and advanced features like native vector search and pluggable storage engines.

  • Project
    oceanbase/seekdb

    A MySQL-compatible state store for AI agents featuring hybrid vector and full-text search, high-concurrency streaming writes, and copy-on-write sandboxes for safe state exploration.

  • Project
    1yefuwang1/vectorlite

    A runtime-loadable SQLite extension that enables fast, SIMD-accelerated vector search using HNSW, providing a SQL interface for approximate nearest neighbor queries.

  • Project
    lancedb/lancedb

    A multimodal AI lakehouse and vector database built on the Lance columnar format for fast, scalable search over petabytes of vectors and multimodal data.