Product Roadmap¶
This roadmap outlines planned features that are not yet active in the current release.
Focus¶
- Better scaling and operations for Triton deployments
- Higher productivity in the workspace
- Clearer administration and security options
Planned Enhancements¶
1) Deployment, Operations, and Performance¶
Resource Management for Triton Instances¶
Goal:
- Provide clear management of GPU, RAM, and CPU resources for self-deployed Triton instances during deployment.
- Enable explicit configuration of requests and limits per self-deployed Triton instance.
2) Security and Networking¶
TLS Configuration in the Deployment Flow¶
Today:
- Add Deployment supports ingress host/class configuration.
- HTTPS TLS is managed outside Triton Control (for example via cert-manager resources or pre-created TLS secrets).
Goal:
- Add TLS configuration directly in the Add Deployment UI.
- Support cert-manager (Issuer/ClusterIssuer) and existing TLS secrets.
Ingress Authentication Enhancements¶
Goal:
- Extend ingress configuration with HTTP Basic Authentication support.
- Add support for Bearer token authentication.
Scope:
- Both authentication options apply to ingress endpoints of self-deployed Triton serving instances created via Triton Control.
3) Platform and Workspace¶
OpenAI API Support for Triton Server¶
Goal:
- Provide OpenAI API support as an integration option for Triton server workflows.
User Management with Groups¶
Goal:
- Extend user management with groups.
- Establish a foundation for role- and team-based administration.
Enhanced S3 Browser¶
Goal:
- Enable renaming files in the S3 Browser.
- Enable moving files and folders in the S3 Browser.
Prioritization¶
Next¶
- TLS configuration in the Add Deployment flow
- OpenAI API integration for Triton Server
Later¶
- Resource management for Triton deployments (GPU/RAM/CPU)
- User management with groups
- S3 Browser: rename and move
Delivered¶
- Per-model performance Jobs: independent non-root Jobs, model-scoped status, and cancellation.