Zero-trust networking secures every connected system as though the network itself offered no protection. Security controls may exist at the network layer, but no system’s security posture should depend solely on network location, IP ranges, firewalls, load balancers, or other network settings. section 4.3.2
Use the architecture to shift trust decisions toward independently secured systems:
- Assess and maintain every connected system as secure, whether it runs inside the environment or could be exposed to the public internet. section 4.3.2
- Treat network-layer rules and CNI policies as additional controls rather than the foundation of the security posture. section 4.3.2
- Use platform policy-as-code and a standardized control plane to make trust controls enforceable and portable. section 4.4
Restricting internal APIs
At Epetech, customer-facing APIs coexist with APIs that must only receive requests from other Epetech services. Explicit network rules could restrict IP ranges or pod traffic through the CNI, but relying on those rules would make the network the security boundary. A zero-trust design instead requires each service to be secure independently, allowing safer external exposure and more flexible network scaling. section 4.3.2
meshCongif.outboundTrafficPolicy.mode = REGISTRY_ONLY
Inspect the service-mesh setting that blocks outbound traffic unless the destination is registered. section 4.3.2
