Krishiv Engine
Rust-native compute for batch SQL, streaming, and experimental incremental processing.
Krishiv Engine is a pre-release compute engine built in Rust around Apache Arrow, DataFusion, and Tokio. It gives batch, streaming, and incremental workloads a shared data model and runtime boundary.
Pre-release software
Krishiv has not reached a stable release. Use it for evaluation, prototyping, and development. APIs, storage metadata, and wire protocols can still change.
Choose a path
Run your first query
Build from source and execute SQL in embedded mode.
Understand the engine
See how sessions, planning, runtimes, and workers fit together.
Check feature maturity
Separate available paths from previews and experiments.
Deploy one node
Start a local coordinator, executor, and Flight SQL endpoint.
What the engine owns
- Batch SQL over registered Arrow and Parquet-style sources.
- Streaming execution with event-time windows, watermarks, and stateful operators.
- Incremental processing through weighted change batches and
IncrementalFlow. - Runtime placement from an embedded process to explicit remote coordinator/executor paths.
- Data-system primitives for shuffle, state, checkpoints, connectors, and observability.
Krishiv Engine does not require Krishiv Platform. The upcoming Platform product will consume Engine through its public APIs and protocols.
How to read these docs
Every page carries a maturity label. “Available” means the path exists in the current source tree; it does not mean the project is stable or production-ready. Preview and experimental paths need extra validation before you rely on them.