Run complex analytical queries with GROUP BY and aggregations about 200x faster than InnoDB by creating tables with the DuckDB columnar engine.
Store AI embeddings alongside regular relational data and do fast semantic similarity search using AliSQL's built-in HNSW vector index.
Replace MySQL 8.0 with AliSQL as a drop-in to gain analytics and vector capabilities without rewriting application queries or changing drivers.
Mix transactional InnoDB tables and analytical DuckDB tables in the same database, choosing the storage engine per table at creation time.
| alibaba/alisql | 0xz0f/z0fcourse_reverseengineering | landrop/landrop | |
|---|---|---|---|
| Stars | 5,812 | 5,821 | 5,831 |
| Language | C++ | C++ | C++ |
| Setup difficulty | hard | easy | easy |
| Complexity | 4/5 | 4/5 | 1/5 |
| Audience | developer | developer | general |
Figures from each repo's GitHub metadata at analysis time.
Must be compiled from source using CMake, Python 3, and GCC or Clang, no prebuilt binaries are mentioned in the README.
AliSQL is a modified version of MySQL, the widely used open-source database, built and maintained by Alibaba. It is based on MySQL 8.0 and is fully compatible with standard MySQL tools, drivers, and SQL syntax, meaning you can use it as a drop-in replacement without changing how you write queries or connect your applications. The main additions over standard MySQL are two: a built-in analytics engine powered by DuckDB, and native support for vector search. The DuckDB engine uses a columnar storage format that is designed for running complex analytical queries across large datasets, which is a different kind of workload than typical transactional queries that read and write individual rows. The README claims around 200 times faster performance on analytical queries compared to the standard InnoDB storage engine. You pick the engine per table when you create it, so transactional tables and analytical tables can coexist in the same database. The vector search feature lets you store high-dimensional numerical arrays alongside regular data and then search for the closest matches by mathematical distance. This is the kind of operation used in AI applications when you want to find content that is semantically similar to a query, rather than finding exact keyword matches. AliSQL supports vectors with up to 16,383 dimensions and includes an indexing method called HNSW for fast approximate nearest-neighbor lookups. AliSQL was open-sourced by the Alibaba Cloud Database Team in December 2025. Building it from source requires CMake, Python 3, and either GCC or Clang. Planned future additions include faster schema changes, quicker crash recovery, and replication improvements. Full documentation and setup guides are included in the repository's wiki directory.
A MySQL 8.0 fork from Alibaba that adds a built-in analytics engine powered by DuckDB (about 200x faster for complex queries) and native vector search, fully compatible with existing MySQL drivers and SQL syntax as a drop-in replacement.
Mainly C++. The stack also includes C++, MySQL, DuckDB.
Setup difficulty is rated hard, with roughly 1day+ to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.