Trending:
Software Development

Alibaba resurfaces stalled AliSQL with DuckDB, vector search additions

Alibaba's MySQL fork AliSQL showed signs of life after six years of dormancy, with commits adding DuckDB engine integration and vector search capabilities. The project last shipped releases in 2018. Whether this signals genuine revival or experimental tinkering remains unclear.

Alibaba's AliSQL project received its first meaningful commits in years, adding DuckDB engine support and vector search integration to the MySQL 5.6 fork. The repository shows commits from January 2026 after effectively going dark since 2018.

AliSQL was open-sourced in August 2016 as Alibaba's production MySQL variant, built to handle events like Singles Day (Double 11). The project contributed 41 features and 40+ bug fixes upstream to MySQL and MariaDB, including hot row contention handling and threadpool optimizations. It powers Taobao, Tmall, and Alipay's backend infrastructure.

The recent additions target modern analytics workloads. DuckDB integration would theoretically allow OLAP queries alongside MySQL's OLTP operations in a hybrid setup. Vector search capabilities align with semantic search and embedding storage trends, competing with dedicated solutions like Milvus and Qdrant.

The problem: AliSQL is based on MySQL 5.6, a version that reached end-of-life in February 2021. Mainstream deployments moved to MySQL 8.0 years ago. The repository shows 5,000 stars and 861 forks but no sustained development activity between 2018 and 2026.

Alibaba evolved its production systems to X-Cluster, a MySQL 5.7-compatible distributed database using X-Paxos consensus for geo-redundancy. That suggests AliSQL itself isn't running Alibaba's current infrastructure.

The timing raises questions. DuckDB integration makes sense for organizations wanting columnar analytics without deploying separate OLAP databases like ClickHouse or StarRocks. Vector search integration targets the AI/ML workload boom. But bolting these onto a seven-year-old MySQL fork seems like an odd foundation.

Organizations evaluating this should consider: Is Alibaba staffing this for real, or is this a research project that will disappear again? Who's maintaining security patches for the ancient MySQL base? What's the migration path if you adopt this and it stalls?

The code exists. The commits are real. Whether this becomes a maintained project or another abandoned experiment, we'll know by whether the next commit comes in weeks or years.