Apache Samza vs Feldera

Apache Samza is a free, open-source distributed framework for applications that process data from multiple sources in real time. It supports stateless and stateful workloads, using a scalable, fault-tolerant state store for retained application state. Developers can use the high-level Streams API, lower-level Task API, Samza SQL or an Apache Beam API; Beam support for Python and Go is described as work in progress. Java and Scala applications can embed Samza as a lightweight client library. Built-in integrations include Apache Kafka, AWS Kinesis, Azure Event Hubs, Elasticsearch and Apache Hadoop, with custom sources also supported. Samza runs on YARN, Kubernetes or as a standalone library across public-cloud, containerized and bare-metal environments. Host affinity and incremental checkpointing help recover tasks and their state after failures. Processing is at least once. The project supports SQL processing and event-time windows, and its documentation describes applications with several terabytes of state and thousands of cores. Samza is self-hosted on Linux and distributed as source artifacts and through Maven; there is no binary release at this time.

Feldera is an incremental view maintenance engine that keeps SQL views current as input data changes, for use alongside a lakehouse or warehouse. Instead of rebuilding a full dataset after each change, it updates views in proportion to the change. It supports SQL window functions and recursive queries. 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, with several output connectors marked experimental. The free open-source edition runs as a single node and container, supports all connectors and datasets larger than memory, and includes community support through Slack. Docker is available for local development, testing, and prototyping, alongside a free online sandbox. Enterprise runs on a customer-provided cloud or Kubernetes cluster and adds multi-node deployments, fault tolerance, resource isolation, and dedicated support. Its documentation says data stays on the customer's premises. Enterprise pricing is custom, and pipeline scale-out is listed as coming soon.