Bolt Backend Features
The Bolt backend integrates Gluten with ByteDance Bolt, a native execution engine for analytical workloads. Bolt has been developed with production Spark workloads in mind, including concerns such as native execution performance and runtime stability.
The goal of the Gluten Bolt backend is to make those capabilities available through Gluten’s backend abstraction. Users can evaluate Bolt through the same Spark plugin model while using Gluten documentation for build, configuration, compatibility, migration, and validation steps.
This page is a navigation entry point. It does not duplicate Bolt’s feature articles, redefine Bolt feature categories, or turn discussion notes into formal performance claims.
Feature Overview
Bolt feature and optimization details are maintained by the Bolt project. This page intentionally does not duplicate those articles or redefine Bolt feature categories in Gluten documentation.
For the latest detailed discussion of Bolt-specific features and optimizations, refer to the Bolt documentation:
Gluten Integration References
The following Gluten documents cover how to build, enable, configure, and validate the Bolt backend from the Spark side.
| Topic | Gluten documentation |
|---|---|
| Build and smoke test | Bolt Quick Start |
| Migration from Velox | Velox to Bolt Migration Guide |
| Backend configuration | Bolt Backend Configuration |
| Spark configuration compatibility | Bolt Spark Configuration |
| Function compatibility | Scalar, Aggregate, Window, Generator |