Feldera

Overview

Feldera maintains SQL views as input data changes, for teams that need continuously updated results alongside a lakehouse or warehouse. Its strongest case is incremental SQL processing; production teams needing distributed deployment and dedicated support must choose the custom-priced enterprise edition.

It supports stateful processing and event-time windows, and SQL includes window functions and recursive queries. Feldera cites complex pipelines handling millions of changes per second, including on a laptop; that claim makes it worth evaluating, but workload requirements should determine whether it fits a particular deployment.

Key features

Incremental SQL processing

Rather than rebuilding a view from an entire dataset after each change, Feldera updates it in proportion to the input change. That is useful when continuously changing data would make repeated full recomputation costly, while support for window functions and recursive queries accommodates more involved SQL transformations.

Connectors and deployment

Documented inputs include Kafka, Google Pub/Sub, NATS, PostgreSQL, AWS S3, Delta Lake, and Apache Iceberg. Outputs include Kafka, Delta Lake, PostgreSQL, Redis, DynamoDB, Snowflake, and Iceberg; several output connectors are experimental. Teams relying on one of those outputs should account for that status before using it in a production design.

Docker supports local development, testing, and prototyping, and a free online sandbox provides another way to explore Feldera. The open-source edition is single-node and single-container; production enterprise deployments run on customer-provided cloud or Kubernetes infrastructure. Feldera says enterprise data never leaves the customer's premises.

Enterprise and security

Enterprise adds multi-node clusters, pipeline and control-plane fault tolerance, resource isolation, multi-tenant isolation, and dedicated support. Pipeline scale-out is coming soon, so teams should not treat it as an existing capability. The Trust Center lists SOC 2 Type 1 and Type 2 compliance and offers security documentation, including penetration-test and SOC 2 reports.

Pricing

Open source edition: free. It includes a single node in a single container, full SQL support, all connectors, datasets larger than memory, and community support. This is a substantial way to build and evaluate a single-node setup, but it does not include enterprise multi-node deployment, isolation, or dedicated support.

Enterprise edition: custom pricing. It is self-hosted on your Kubernetes clusters and includes multi-node deployments, multi-tenant isolation, and dedicated support. This is the relevant tier for production teams that need those capabilities; its custom pricing means the free edition is the only option with a stated price.

Community support for the open-source edition is through Slack. Feldera's enterprise offering also includes dedicated support.

Platforms

Feldera lists API, Linux, macOS, self-hosted, web, and Windows platforms. Its deployment model is hybrid: local Docker and an online sandbox serve development and exploration, while enterprise production runs in the customer's cloud or Kubernetes cluster.

Who it's for

Feldera suits data teams that want SQL views to stay current as source tables change, particularly for real-time medallion architectures, fraud-detection feature engineering, or fine-grained authorization. The free single-node edition can suit evaluation and smaller deployments; teams requiring multi-node production, fault tolerance, resource isolation, or dedicated support need enterprise. It is less suitable when a required output depends on an experimental connector or when pipeline scale-out is a prerequisite.

Pros and cons

• Incremental view updates: changes are applied in proportion to input changes, avoiding full-dataset recomputation for every update.
• Broad SQL expressiveness: window functions and recursive queries support more than basic transformations.
• Useful free-edition scope: full SQL, all connectors, and datasets larger than memory are included, though deployment remains single-node and single-container.
• Customer-controlled enterprise hosting: Kubernetes or cloud deployment keeps data on the customer's premises, with multi-node and isolation capabilities.
• Connector caveat: several outputs are experimental, which makes them a less certain foundation for production dependencies.
• Enterprise trade-off: multi-node deployment and dedicated support require custom pricing, and pipeline scale-out is still coming soon.

Alternatives

For a free, open-source streaming framework, consider Apache Storm. Apache Beam is a free open-source programming model, with execution costs dependent on the runner and infrastructure. Arroyo offers a free, Apache 2.0-licensed self-hosted engine as a single binary.

Kafka Streams is a free option available for download, without formal or commercial support. Timeplus has a free Apache 2.0 open-source streaming SQL engine as well as a free trial. Apache NiFi is free to download under Apache License 2.0. Apache Spark is a free open-source distributed data analytics engine with download, PyPI, Maven Central, and Docker options. AFL++ is a free tool offered under AGPL terms.

Browse Stream Processing Software and Streaming Analytics Software for more options by category.

Verdict

Choose Feldera if your team wants continuously maintained SQL views and can start with a single-node edition or justify custom-priced enterprise deployment. Incremental updates, broad SQL support, and a free edition make it a focused fit for changing data pipelines. Look elsewhere if you require established production support for an experimental output connector or immediate pipeline scale-out.