Skip to content

fhir · healthcare

FHIR integration on EKS

How a minimal FHIR server runs inside a Kubernetes cluster under real healthcare compliance — what the request path looks like, and where the auditability lives.

Author
Lali Devamanthri
Published
Reading time
1 min read

Placeholder article. Real content to follow. The structure below is real — it's the shape every FHIR-on-Kubernetes piece will take — but the words are scaffolding.

FHIR is what healthcare integration looks like when it's well-engineered. The spec is large, but the runtime shape of a production FHIR server is small and opinionated — an ingress surface, a stateless server, a regional database, and a durable audit trail. Getting it wrong means failing an audit.

The diagram below is the default shape I reach for on a regulated EKS cluster.

A healthcare integration layout. A client calls an ingress proxy inside EKS. The proxy routes to a FHIR server pod. The FHIR server reads from a regional PostgreSQL database and forwards audit events to a durable log bucket.

FHIR-on-EKS: ingress terminates at a proxy, routes to a FHIR server pod, which reads from a regional PostgreSQL and forwards audit events to a durable log.ClientEHR · appIngressmTLS · rate-limitFHIR serverEKS podPostgresHA · regionalAudit logappend-onlyCOMPLIANCE

Every request is authenticated, routed, and auditable.

Edge — mTLS and rate-limits

Every call lands on an ingress proxy that terminates mTLS and enforces per-client rate limits. Authentication is out of the FHIR server's job description.

Server — stateless by design

The FHIR server pods are stateless. Horizontal scaling is a matter of replica count. Persistent state lives in the regional Postgres with HA failover.

Audit — append-only, always

Every state-changing request writes to an append-only audit log before it's considered complete. This is the line between a system that can pass an audit and one that can't.

The hard part isn't the FHIR server. The hard part is making the whole environment around it — identity, network policy, audit retention, backup RPO — live up to the integration spec.

End of article

Building something AI-shaped for healthcare or fintech?

I work with a small number of teams at a time on integration architecture, eval pipelines, and getting models into regulated production. If the system you're designing rhymes with the one above, let's talk.