Runtime Plane Deployment Options¶
Overview¶
A runtime plane can be deployed in two ways. They differ in where the control plane runs and who operates it — the connectors always run in your own cloud account either way.
- Partially Managed (D2P) — Trust3AI runs the control plane in Trust3AICloud; only the Runtime Agent and connectors run in your cloud account.
- Self-Managed — the entire stack, including the control plane, runs inside your own cloud account.
You choose the deployment type when you create the runtime plane in the portal (Settings → Runtime Plane). The portal guides the rest and generates the exact install commands for your cluster, so the sections below stay at an overview level.
Deployment types¶
The control plane — Runtime Manager, Policy Manager, Audit Server, and Portal — stays in Trust3AICloud. Only the Runtime Agent and the connector workloads run in your cloud account, in a Kubernetes namespace you provide. Trust3AI operates and updates the control plane; you operate the runtime plane in your VPC. The agent polls the control plane for configuration and pushes status back; your data and connector credentials stay in your account.

- Runs in your account: Runtime Agent, connectors, UserSync, Dataserver, Discovery, engine plugins.
- Runs in Trust3AICloud: Runtime Manager, Policy Manager, Audit Server, Portal.
- Cloud providers: AWS (EKS), Azure (AKS), or Google Cloud (GKE).
- You provide: a Kubernetes cluster and namespace.
- Provisioning: the portal generates a values download and a Helm command that installs the agent into your namespace.
The entire stack runs inside your own cloud account: the control plane (Runtime Manager, Policy Manager, Audit Server, Portal) and the Runtime Agent and connectors. You also provide and operate the surrounding infrastructure. Trust3AI supplies the charts and configuration; you deploy and run everything, so nothing leaves your network.

- Runs in your account: the control plane plus the Runtime Agent, connectors, UserSync, Dataserver, Discovery, and engine plugins.
- Cloud provider: AWS (Amazon EKS).
- You provide: the EKS cluster plus S3, EFS, a database, Route 53 DNS, ALB/ACM ingress, and an IAM role.
- Provisioning: the portal wizard collects your infrastructure, storage and connectivity, and cloud permissions, then generates the values and platform-bundle downloads and a Helm command. After the pods are running, you open your own in-cluster portal.
Comparison¶
| Partially Managed (D2P) | Self-Managed | |
|---|---|---|
| Control plane location | Trust3AICloud | Your cloud account (EKS) |
| Runtime Agent & connectors | Your cloud account | Your cloud account |
| Who operates the control plane | Trust3AI | You |
| Cloud providers | AWS, Azure, Google Cloud | AWS (EKS) |
| Infrastructure you provide | Kubernetes cluster + namespace | EKS + S3, EFS, database, Route 53, ALB/ACM, IAM |
| Control-plane updates & patching | Trust3AI | You |
| Data & credential residency | Your cloud account | Your cloud account |
| Operational overhead | Lower | Higher |
| Portal setup steps | Infrastructure → Deploy | Infrastructure → Storage & Connectivity → Cloud Permissions → Deploy |
Recommendation¶
Use Partially Managed (D2P) unless you have a specific reason not to. Your connectors and data stay in your cloud account while Trust3AI runs and maintains the control plane, which keeps operational overhead low.
Choose Self-Managed when policy requires the entire stack — control plane included — to run inside your own network.
- Prev topic: Runtime Plane
- Next topic: Before You Begin