Zero-Trust Architecture: Micro-segmentation and Identity-Driven Access

Traditional perimeter security assumes that everything inside the corporate network or VPC is trusted. In zero-trust architecture, the core principle is simple: never trust, always verify.

1. Cryptographic Workload Identity

Rather than relying on IP addresses or firewall rules for access control, zero-trust relies on cryptographically verifiable workload identities:

• SPIFFE / SPIRE: Issues short-lived X.509 SVID certificates to workloads based on verifiable platform attributes.
• Mutual TLS (mTLS): Enforces bidirectional encryption and identity attestation between microservices.

2. Dynamic Authorization via Policy Engines

Moving access control logic out of application code into centralized policy engines like Open Policy Agent (OPA) or Cedar:

package authz

default allow = false

# Allow read access to patient records if authenticated with doctor role
allow {
    input.method == "GET"
    input.path = ["v1", "records", _]
    input.jwt.claims.roles[_] == "physician"
}