Note
Access to this page requires authorization. You can try signing in or changing directories.
Access to this page requires authorization. You can try changing directories.
MLflow tracing helps you monitor GenAI applications in production by capturing execution details. You can deploy traced applications in two ways:
- On Databricks: Deploy using Agent Framework or custom model serving with full integration for monitoring and inference tables
- Outside Databricks: Deploy to external environments while logging traces back to Databricks for monitoring
Compare deployment options
The table below compares features available for each deployment location:
| Deployment location | MLflow experiment trace logging | Production monitoring | Inference tables |
|---|---|---|---|
| Deploy on Databricks | Supported | Supported | Supported |
| Deploy outside Databricks | Supported | Supported | Not supported |
- MLflow experiment trace logging: Real-time trace logging with MLflow experiment UI access. View traces in the UI or query programmatically. Supports up to 100K traces per experiment.
- Production monitoring: Automatic archival to Delta tables for long-term analysis and monitoring. Adds ~15 minute delay compared to experiment logging.
- Inference tables: Available only for Databricks deployments. Stores traces in Delta tables with no experiment limits. Adds 30-90 minute delay and has trace size limits.
Next steps
Choose your deployment approach: