Realfuture Documentation
Realfuture is a one-stop AI-driven streaming lakehouse platform for data integration, development and operations across your infrastructure. Shared catalogs connect your data to SQL, CDC and ETL development; AI assists with authoring and diagnosis; one platform manages jobs across multiple clusters.
Built on Apache StreamPark by its original creator, Realfuture adds enterprise capabilities across the platform and its data ecosystem, including maintained Flink, Flink CDC and connector enhancements.
Work with Flink for processing, Flink CDC for continuous data synchronization, Paimon for lake tables, Fluss for streaming storage, and Doris or StarRocks for analytics. Realfuture brings their metadata, development and operating workflows together.
Understand the platform
| Capability | What you can do |
|---|---|
| Catalogs | Discover databases, streams and lakehouse tables; reuse schemas, views and functions in development |
| AI-assisted development | Generate SQL, CDC configuration and ETL definitions with schema context; explain errors and review suggestions |
| Data integration | Synchronize data with Flink SQL or Flink CDC pipelines; configure table routing, filtering, field mapping and transformations, with whole-database replication and schema evolution for supported CDC paths |
| Lakehouse workflows | Develop against Paimon and Fluss tables and connect downstream analytics in Doris or StarRocks |
| Queries and validation | Check SQL and configuration, preview results and query supported databases through Native SQL |
| Production operations | Publish jobs, inspect runtime evidence, use AI diagnosis and manage recovery with workspace controls |
| Multi-cluster deployment | Deploy and operate Flink jobs across multiple YARN and Kubernetes clusters from one Realfuture installation |
One platform across your clusters
A single Realfuture deployment can manage jobs on separate YARN and Kubernetes clusters, including clusters in different cloud accounts or data centers. For example, teams can use the same platform to deploy jobs to Kubernetes on AWS and Google Cloud while continuing to operate jobs on an existing YARN cluster.
Each target keeps its own compute, access and storage configuration. Register the environments, configure credentials and verify the required network paths before publishing. See deployment options and compatibility.
Choose the next guide
- Explore the product overview for the platform capabilities and their relationship.
- Start with catalogs and metadata, then development, AI assistance and publishing.
- Read streaming lakehouse development for component roles and table maintenance.
- Explore data integration and team workflows.
- Review deployment options, compatibility and software and support.
Evaluate a complete workflow
Choose a representative source, target and business question. Connect its catalog, develop and validate a pipeline, inspect target data, then test a schema change or runtime failure. Include AI generation and diagnosis when these are part of your requirements.
AI uses configured model providers and scoped context. Users review generated changes before applying them; diagnosis gathers read-only evidence. Confirm model access, data handling and edition entitlements during evaluation.
Use the evaluation guide to agree on success criteria, or book a demo. Installation artifacts and instructions must match the agreed product release and deployment.