Truefoundry Control Plane - TrueFoundry Docs

The control-plane is the brain of the Truefoundry platform. It is responsible for managing the overall configuration and orchestration of the platform components.

We provide a hosted control-plane that you can use or also self-host the control-plane as part of our enterprise plan

The control-plane comprises of the following key components. Different components are required for different use-cases. If you are not using those features, they can be disabled and will not be installed.

Component	AI Deployment	AI Gateway	Description
UI			This is the UI component to view deployments, and other resources.
Backend Microservice			Truefoundry comprises of multiple backend services that handle various aspects like authorization, deployment control flow, CRUD APIs, interaction with database and external services, etc.
PostgreSQL			Database to store user information, deployment information, etc, gateway configuration, etc. This can be deployed on Kubernetes in dev-mode, however, we recommend using managed database like RDS in production.
Blob Storage			Control-plane needs access to a blob storage to store code uploads, gateway logs and traces.
Queue			Truefoundry uses NATS as the queue and cache layer to communicate all data from the control plane to the compute and gateway planes.
Otel Collector			Otel collector to collect tracing data from AI Gateway
Ingestor			Ingestor receives the logs from the OTEL collector and dumps them to S3 or any other blob storage.
Controller			The controller is responsible for handling all connections from the tfy-agents in different compute-plane clusters.
Image Builder			ImageBuilder is used in AI deployment to build docker images on Kubernetes cluster using buildkit.
Workflows Microservice			Workflows microservice is used in AI deployment to manage Truefoundry workflows. This is needed only if you are using workflows feature.
Spark History Server			Spark history server is used in AI deployment to view the logs and traces of the spark jobs deployed on Truefoundry . This is needed only if you are using deploying Spark jobs on Truefoundry.

Control-plane also integrates with a number of external services to make it easier to manage your deployments and configuration. These integrations are optional and you can only enable the ones you need. The key types of integrations and their usecases are as follows:

Integration Type	AI Deployment	AI Gateway	Description
Docker Registry			Needed to pull images for deployed services. Required for building and deploying AI services.
Blob Storage			Used to store code uploads, model artifacts, logs, and gateway traces.
External Secrets Manager (e.g., AWS Secrets Manager, GCP Secret Manager)			Used to securely inject sensitive credentials into services.
Authentication System (OIDC, SSO)			Provides user authentication and single sign-on capabilities for the platform.
Notifications (Slack, Email, Webhooks)			Allows configuring notifications for deployment status, failure alerts, and other platform events.
Git Repository Integration			Enables seamless integration with Git repositories for CI/CD workflows, code synchronization, and automated deployment triggers.

Supported Integrations

Docker Registry
Blob Storage
Secret Manager
Notifications
Git Repositories

AWS Container Registry

GCP Artifact Registry

Azure Container Registry

Docker Hub

Quay

JFrog Artifactory

Self-hosted registries

​Supported Integrations

Supported Integrations