Why Fluss Is the Next Big Leap in Real‑Time Stream Storage

On June 5, the Fluss project—Alibaba's open‑source next‑generation stream storage—successfully passed the vote and became an Apache Software Foundation (ASF) incubator project, marking a milestone toward a more open, neutral, and standardized development stage.

The Fluss community has completed the donation process and officially transferred the project to the ASF. During the Flink Forward Asia 2025 keynote in Singapore on July 3, project initiator Wu Chong announced the new repository (https://github.com/apache/fluss/) and official website (https://fluss.apache.org/).

Fluss is a next‑generation stream storage engine designed for real‑time analytics scenarios, addressing high cost and low efficiency in traditional stream storage for stream computing and lakehouse use cases. Its core features include:

Columnar stream storage: Millisecond‑level latency for real‑time read/write, using Apache Arrow columnar format with column and partition pruning to achieve up to 10× read performance and lower network costs.

Real‑time updates and point queries: Innovative real‑time update capabilities integrated into stream storage, enabling high‑performance streaming updates, partial column updates, binlog, dimension table point queries, and DeltaJoin for low‑cost real‑time data warehouses.

Lake‑flow integration: Unified lake and stream storage for data sharing, providing low‑cost historical data support to lakehouse and injecting real‑time capabilities, achieving a seamless lake‑warehouse experience.

Fluss originated in July 2023 when Alibaba Cloud's intelligent Flink team launched the project, named from “Flink Unified Streaming Storage”. The name also means “river” in German, symbolizing continuous data flow.

After more than a year of internal incubation, the project was officially announced at the Flink Forward Asia 2024 conference in Shanghai on November 29, attracting over 60 global contributors and releasing a major version roughly every three months.

Within Alibaba, Fluss now supports over 3 PB of data, with peak cluster throughput of 40 GB/s, single‑table point‑query QPS of 500 k, and tables up to 500 billion rows, delivering strong performance in log analysis, search recommendation, and real‑time data warehousing.

Joining the ASF incubator aligns with Fluss's goals of openness, collaboration, and neutrality, accelerates integration with other Apache projects (e.g., Flink, Spark, Kafka), and provides robust governance and sustainable development support.

The community thanks its mentors and contributors, including PMC members from Flink, HBase, Pulsar, ZooKeeper, and other Apache projects.

Developers and users are invited to join the Fluss community, follow updates, and contribute via the GitHub repository (https://github.com/apache/fluss/), official website (https://fluss.apache.org/), mailing list ([email protected]), and community chat groups.