Apache Fluss is an open-source, lakehouse-native streaming storage system. It enables Lakestream: a shared table foundation coordinating fresh streaming data and historical lakehouse data for the Streamhouse architecture.
-- Register Apache Fluss as a Flink catalogCREATE CATALOG fluss_catalog WITH ('type' = 'fluss','bootstrap.servers' = 'coordinator-server:9123');USE CATALOG fluss_catalog;-- Create a primary-key tableCREATE TABLE pk_table (shop_id BIGINT,user_id BIGINT,num_orders INT,PRIMARY KEY (shop_id, user_id) NOT ENFORCED) WITH ('bucket.num' = '4');INSERT INTO pk_table VALUES (1234, 1234, 1);SELECT * FROM pk_table WHERE shop_id = 1234;
Streamhouse brings streaming, serving, and analytics together through shared tables and independent compute engines. Fluss enables its Lakestream storage foundation by coordinating fresh streaming data with historical lakehouse data.
Explore Streamhouse and LakestreamA common stream can feed several systems that each ingest, reconstruct, and synchronize equivalent data. Streamhouse lets compatible workloads reuse maintained tables on a Lakestream foundation. Compute engines still run transformations; specialized stores remain useful when a workload needs them.
Repeated ingestion · Equivalent state · Synchronization work
Shared tables · Managed tiering · Compatible access paths
Apache Fluss enables shared table maintenance and access across streaming and lakehouse storage, with independent engines providing computation, queries, and serving.
Reusable tables for streaming, serving, and analytics.
Independent engines and applications share maintained datasets through supported interfaces, reducing repeated ingestion and reconstruction of equivalent data.
Architectural basisShared table schemas, row semantics, and supported access paths.
Fresh and historical data as one logical table.
Streaming and lakehouse representations remain coordinated through shared metadata, managed tiering, and supported reads across their different freshness layers.
Architectural basisTable metadata, committed tiering progress, and Union Read integrations.
Independent engines operating on shared tables.
Engines run transformations and publish reusable result tables. Private execution state, including windows and timers, remains the responsibility of each computation.
Architectural basisSeparate compute and storage services with supported table interfaces.
Pruning that compounds.
Server-side projection, predicate pushdown, and partition pruning on Arrow-format streams compound into order-of-magnitude I/O and network savings.
Architectural basisARROW log format and the compound pruning stack on the TabletServer.
Maintained data for ML serving and AI context.
Applications retrieve shared features and context through supported interfaces while owning their retrieval policies, decision logs, and any specialized indexes.
Architectural basisPrimary-key lookups, streaming reads, and lake format integrations.
Documented formats and supported APIs.
Compatible engines read streams, committed lake data, or both through a supported union integration. Capabilities depend on the engine and lake format.
Architectural basisStreaming APIs, open lake formats, and catalog integrations.
Kafka is the streaming transport. Fluss is the streaming storage. If your need is large-scale stream processing with Flink, real-time analytics, AI/ML, or a sub-second lakehouse, Fluss is the shared streaming storage substrate behind all of them. Read the full breakdown to see which fits your stack.
Apache Fluss is developed openly by a global community of contributors. Join the discussion, file an issue, or send a patch.
View on GitHub