Timeplus Proton

Timeplus is an open-source SQL pipeline engine for stream processing, analytics, and observability. It allows users to ingest live data, perform transformations, and route information to downstream systems. The software is deployed as a single C++ binary or via Docker, requiring no JVM or ZooKeeper dependencies, which simplifies the deployment process for data engineers.

Users interact with the system primarily through a SQL interface to create external streams and tables. It supports common data engineering tasks such as streaming ETL, real-time telemetry processing, and the creation of feature pipelines for AI and machine learning. The system is designed to be resource efficient and can run on minimal hardware, including small cloud instances with as little as one vCPU and 0.5 GiB of memory.

Key features

• Native source and sink support for Kafka, ClickHouse, MySQL, Postgres, MongoDB, and S3/Iceberg
• Incremental materialized views for real-time analytics and dashboards
• Support for tumble, hop, and session windows with watermarks
• User-defined functions written in Python or JavaScript
• Integration with Kafka and Redpanda via external streams
• High-throughput processing utilizing SIMD optimizations
• Built-in support for CDC and data revision processing
• SQL-based alerting and task management

Built on the ClickHouse engine, Timeplus extends historical data storage and computing functionality with real-time stream processing capabilities. It provides SDKs for Python, Java, and Go, as well as a REST API for integration into existing application stacks. The architecture is designed for low-latency requirements, enabling high-cardinality aggregations and millisecond query responses over billions of rows. This makes it suitable for building real-time telemetry pipelines that reduce noise before forwarding logs and metrics to stores.

Timeplus is a lightweight, single-binary stream processing framework.