Skip to content

Deployment Model Notes

Deployment Model Notes

Deployment-model notes for dedicated project runtimes and client-owned boundaries.

This page summarizes the deployment assumptions behind the Gerimedica architecture views.

What is dedicated

Each Gerimedica project is expected to resolve to a dedicated C2F runtime with:

its own API surface
its own relational state
its own search/index state
its own secret boundary
its own provider project and cluster in the production model

What remains client-owned

Gerimedica remains the owner of:

the connector runtime that submits PHI-minimized records
the answer endpoint used by the Gerimedica-managed agent
durable answer storage
UUID-to-patient mapping
zorginstelling-level access control

Requirements

Outbound access from the connector to the batch API
Reachability from the C2F agent proxy to the client answer endpoint
A provider environment capable of hosting a dedicated Kubernetes cluster and project-local data services
Operational readiness for project-level suspend, resume, throttle, or scale controls

Provider-specific layouts

See the provider-specific diagrams for the concrete runtime shape:

Spend-control lens

For the operating model used to track spend per client project and expose suspend/resume hooks, see:

Spend Control