featzhang opened a new pull request, #27490:
URL: https://github.com/apache/flink/pull/27490
### What is the purpose of the change
This PR adds comprehensive documentation for the **Triton inference model**
in Flink SQL, enabling users to perform real-time model inference by calling
**NVIDIA Triton Inference Server** from Flink.
The documentation introduces how to define, configure, and use Triton-backed
models in Flink SQL, covering common inference scenarios as well as advanced
production use cases.
### What is the change
* Add a new **Triton model documentation** page under
`docs/connectors/models`
* Overview of Triton integration and supported features
* End-to-end SQL examples for model creation and inference
* Advanced configurations (authentication, headers, batching, compression,
priority)
* Examples for array-type inference and stateful / sequence models
* Add corresponding **Chinese documentation** to keep EN / ZH docs consistent
* Extend existing SQL documentation to reference Triton models:
* `CREATE MODEL` syntax
* `ML_PREDICT` usage for model inference
### Why is the change needed
Triton is a widely used, high-performance inference serving system
supporting multiple ML frameworks.
Providing first-class documentation for Triton model integration helps users:
* Integrate online model inference into Flink SQL pipelines more easily
* Understand supported configuration options and best practices
* Adopt Flink as a unified platform for real-time data processing and AI
inference
### How was this change tested
* Documentation-only change
* Verified SQL examples and configuration options for correctness and
consistency
### Does this PR introduce any user-facing change
Yes.
This PR introduces new user-facing documentation describing how to use
Triton inference models in Flink SQL.
### Checklist
- [x] Documentation updated (English and Chinese)
- [x] SQL examples provided
- [x] No backward compatibility impact
- [x] No code or runtime behavior changes
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
