1 of 10

Protocol Gap Analysis for Zero Trust

Control and Policy Interoperability

Overview for IETF ZTCPP

Why Zero Trust still lacks interoperable connectivity control-plane semantics

POLICY

PLANE

CONTROL

PLANE

SESSION /

FLOW

Identity-agnostic, service-centric interoperability for human and non-human connectivity

IETF ZTCPP Gap Analysis

Philip Griffiths

2 of 10

The structural flaw

Traditional TCP/IP networking establishes reachability before identity and policy are evaluated.

If a service is addressable, it is discoverable and reachable before authorization.

This connect-then-authorize ordering creates systemic exposure that Zero Trust is supposed to reduce.

Connect first, authorize later is the wrong default — across users, workloads, and services.

The paper’s architectural thesis

IETF ZTCPP Gap Analysis

Philip Griffiths

3 of 10

3

What this paper is - and is not

A CSA/NIST-aligned framing for the IETF ZTCPP effort

This paper IS

  • A gap analysis for interoperable connectivity control-plane protocols.
  • Explicitly aligned to broader Zero Trust principles (CSA, NIST SP 800-207).
  • Applicable across in-network, gateway-centric, and overlay/service-centric models.
  • Identity-agnostic: composes with existing IdP, PKI, and workload identity systems

Connectivity control plane = how policy is expressed, distributed, updated, revoked, and bound to sessions/flows with verifiable enforcement.

This paper is NOT

  • A new wire protocol specification.
  • A prescription for a single Zero Trust architecture.
  • A standard for endpoint security products (EDR/MDM) themselves.
  • Not limited to user access (applies to services, workloads, agents, and APIs)

But posture, attestation, and context signals — and their protocol interactions — are in scope because dynamic trust is central to operational Zero Trust.

IETF ZTCPP Gap Analysis

Philip Griffiths

4 of 10

The three interoperability seams

Three seams define the control-plane problem.

The seams are distinct because they govern different failure modes:

  • (4.1) exposure failure
  • (4.2) coordination failure
  • (4.3) enforcement fidelity failure

Three distinct interaction surfaces — exposure, control, and runtime enforcement

IETF ZTCPP Gap Analysis

Philip Griffiths

5 of 10

5

What this seam must guarantee

  • No ambient reachability before authorization

  • No discovery or probing of protected services

  • Mediated access via gateway, overlay, or similar enforcement point

What breaks without it

  • Routable services remain scannable and exploitable

  • "Authenticate later" leaves pre-auth exposure in place

  • Exposure reduction alone does not provide lifecycle or runtime policy fidelity

Seam 4.1: Gateway <-> Resource

Identity-bound�overlay / mediated�connectivity

Authenticated-before-connect (ABC) and resource-hiding at the exposure boundary

Exposure reduction matters - but exposure reduction is not the whole control plane.

IETF ZTCPP Gap Analysis

Philip Griffiths

6 of 10

6

Controller

Gateway

Agent

Router

register • advertise capability • push/update policy • reauth/revoke • report status

4.2 pushes lifecycle decisions from controller to distributed EPs

What this seam must guarantee

  • EP registration and capability advertisement

  • Policy distribution, update, revoke, reauth, and state sync

  • Posture/context propagation with predictable convergence

What breaks without it

  • Multi-vendor policy lifecycle fails

  • Dynamic trust signals do not propagate consistently

  • Systems devolve into proprietary adapters and stale state

Policy must move, adapt, and revoke consistently.

Seam 4.2: Controller <-> Enforcement Point

Dynamic trust and policy lifecycle between controller and EPs

IETF ZTCPP Gap Analysis

Philip Griffiths

7 of 10

7

Policy

Decision

Bound Service

Session / Flow

Identity

Service

Context

These inputs must survive contact with the data plane

without dilution into coarse address-based controls.

4.3 is where least privilege becomes observable and verifiable based on strong identity.

What this seam must guarantee

  • Policy intent bound to concrete sessions, flows, or channels

  • Identity-, service-, and context-based access realized precisely at runtime (external identity optional to ensure support any use case) with consistent enforcement across human and non-human initiated sessions

  • Verifiable least privilege at service granularity

What breaks without it

  • Policy can be delivered correctly but enforced too broadly

  • Audit cannot reliably correlate decision to observed traffic

  • Least privilege collapses into coarse address-based enforcement

Least privilege must survive contact with the data plane.

Seam 4.3: Policy Plane <-> Data Plane

Runtime binding of policy intent to concrete sessions and flows

IETF ZTCPP Gap Analysis

Philip Griffiths

8 of 10

8

Key protocol gaps identified in the paper

Section 6 organizes standardization opportunities around seven gaps

G1 Controller <-> EP lifecycle

Problem

No widely adopted interoperable protocol for registration, capability advertisement, policy update, revocation, refresh, and stale-state handling.

Why it matters

Multi-vendor environments require proprietary integrations and inconsistent lifecycle behavior.

G2 Dynamic trust inputs

Problem

Limited interoperability for how posture, attestation, and context signals trigger policy updates, reauth, or enforcement changes across EPs.

Why it matters

Dynamic trust decisions are not portable across implementations.

G3 Decision binding

Problem

No common model linking least-privilege service policy decisions to concrete sessions, flows, or channels.

Why it matters

Audit, revocation assurance, and service-level enforcement verification remain weak or inconsistent.

G4 ABC composition

How gateway/resource ABC/hiding composes with dynamic trust control planes

G5 Service portability

Service- and NHI-based policy targets across address- and service-centric models

G6 Identity / key composition

External IdPs, PKI/HSM, credential ecosystems, workload identity

G7 HA / degraded mode

Failover, stale policy, re-sync, fail-open/fail-closed/fail-limited

The draft’s center of gravity shifts from “ABC only” to service-based least privilege, runtime binding, and portable control-plane semantics.

IETF ZTCPP Gap Analysis

Philip Griffiths

9 of 10

9

Implementation classes compared

Informative comparison: which seams each class exercises most strongly

Implementation class

Strongest seam(s)

Where it helps most

Where it tends to be weaker

Gateway-centric ABC / hiding

4.1

Eliminates pre-auth exposure; strong ingress/egress mediation

Policy lifecycle, service-level least privilege, end-to-end binding

In-network enforcement

4.2

Control distribution and coordinated enforcement across existing fabrics

Portable service semantics; mapping policy intent to runtime behavior

Overlay / service-centric

4.1 + 4.2 + 4.3

Identity-first service access, runtime binding, service-granular policy

Still needs interoperable semantics rather than vendor-specific control models

This comparison is illustrative, not prescriptive: the point is to show why seam coverage matters, not to pick winners.

IETF ZTCPP Gap Analysis

Philip Griffiths

10 of 10

Takeaways for discussion

1. Controller↔EP control protocol requirements

2. Controller↔EP interoperability framework

3. Policy↔data binding framework

4. Dynamic trust / context profile

5. ABC composition profile

6. Service target abstraction

The missing interoperability layer is the one that binds policy intent to real connections.

Discussion:

Which seam is most urgent for interoperable standardization?

Zero Trust is not interoperable until policy can be portably bound to real sessions and flows.

  • Zero Trust is not just “better perimeter filtering”; it requires a structural inversion of connect-before-authorize networking.

  • 4.1 matters, but service-level least privilege only becomes real when 4.2 and 4.3 are also interoperable.

  • The hardest standardization work is not identity credentials alone - it is control-plane lifecycle and runtime enforcement.

  • Open source implementations (e.g., OpenZiti) are valuable because they exercise the seams in real systems and expose where protocol behaviour is still vendor-specific rather than interoperable.

  • The opportunity for ZTCPP is to make service-based least privilege portable, observable, and verifiable across heterogeneous systems.

Why this paper matters for CSA, IETF, and open implementations

IETF ZTCPP Gap Analysis

Philip Griffiths