Timeplus

Key features

• Streaming SQL over Kafka-family sources Proton queries live data in Kafka, Redpanda, and Pulsar through external streams, with native sources and sinks for ClickHouse, MySQL, Postgres, MongoDB, S3 and Iceberg, and OpenSearch.[1][2]

• Windows, watermarks, and CDC handling Tumble, hop, and session windows with watermarks express time-based aggregations in SQL, and CDC plus data-revision processing keeps derived state aligned with changing source rows.[1][2]

• Incremental materialized views in-engine Materialized views are maintained incrementally and stored as queryable analytical views inside Proton itself, so dashboards and applications read results without a separate serving store.[1][2]

• Single-binary C++ runtime Proton ships as one binary under 500 MB with no JVM, ZooKeeper, or other dependencies, and runs on bare metal, edge devices, containers, or a small cloud instance.[1][2]

• Vendor-reported speed example The vendor reports 90 million events per second with 4 millisecond end-to-end latency on an M2 Max MacBook Pro; treat this as a vendor-published example, not an independently verified benchmark.[1][2]

• Python and JavaScript UDFs Custom logic extends streaming SQL through user-defined functions in Python and JavaScript, alongside built-in alerting and scheduled tasks.[1]

• CLI, drivers, and REST APIs Operators and applications connect through a native CLI, JDBC and ODBC drivers, ingest REST and server-sent-events query APIs, and Python, Go, C++, Rust, and OCaml clients.[1]

• Enterprise clustering and SSO The paid Enterprise tier adds multi-node clustering with high availability plus OIDC single sign-on, while single-node Proton covers Docker, binary, and Kubernetes self-hosting.[1][2]

Who will love it

• Kafka-centric streaming teams Teams that already move data through Kafka or Redpanda will value querying live topics in SQL without first landing everything in a warehouse.[1][2]

• Real-time analytics builders Teams building live dashboards and alerting will value in-engine materialized views served through SQL, drivers, and REST query APIs.[1][2]

• Lean-infrastructure operators Teams that want stream processing without a JVM platform will value the dependency-free single binary with script, Docker, and Helm install paths.[1][2]

Who should avoid it

• JVM-platform standard shops Organizations standardized on Flink or ksqlDB operations should think carefully, since Proton is a separate C++ runtime positioned as their alternative rather than a drop-in addition.[1][2]

• Non-technical dashboard users There is no no-code analytics suite; Timeplus expects SQL fluency plus stream, driver, and server configuration from its operators.[1][2]

• Residency-constrained organizations No public data-location or residency options were documented in the sources reviewed, so organizations with such requirements should confirm before adopting.[1][2]

Compatibility