ποΈ Zero Trust Paradigm Shift: "Never Trust, Always Verify"
The legacy enterprise network security postureβoften called the "Castle-and-Moat" modelβrelied on the assumption that anything inside the internal corporate network was inherently trustworthy, while anything outside the firewall was untrusted. Modern adversary capabilities, advanced persistent threats (APTs), hybrid cloud computing, and remote distributed workforces have rendered this physical perimeter obsolete: once an attacker breaches the outer moat (via phishing or an unpatched VPN server), they enjoy uninhibited lateral movement across the corporate intranet. Zero Trust Architecture (ZTA), formally codified in NIST SP 800-207, replaces implicit perimeter trust with an uncompromising engineering philosophy: assume breach, verify explicitly, and enforce least-privilege access continuously. Every single access attemptβwhether originated by an internal database server, an external user, or an ephemeral microserviceβmust be authenticated, authorized, encrypted, and dynamically assessed against continuous contextual signals.
LEGACY PERIMETER MODEL VS. NIST SP 800-207 ZERO TRUST LOGICAL ARCHITECTURE:
1. LEGACY CASTLE-AND-MOAT MODEL (IMPLICIT INTRANET TRUST):
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β UNTRUSTED INTERNET βββΊ [Edge Firewall / VPN Gateway] βββΊ TRUSTED CORPORATE INTRANET β
β βββ Workstation (Unauthenticated) β
β βββ Domain Controller (Wide open) β
β βββ Database (Cleartext traffic) β
β [FLAW: Compromise of 1 VPN account grants lateral movement across entire internal network] β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
2. NIST SP 800-207 ZERO TRUST LOGICAL ARCHITECTURE (SEPARATED PLANES):
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β CONTROL PLANE: POLICY DECISION POINT (PDP) β
β β
β ββββββββββββββββββββββββββββββββ ββββββββββββββββββββββββββββββββ β
β β POLICY ENGINE (PE) βββββββββββββββΊβ POLICY ADMINISTRATOR (PA) β β
β β Evaluates continuous rules, β β Issues/revokes credentials, β β
β β risk scores & posture checks β β commands data plane gates β β
β ββββββββββββββββββββββββββββββββ ββββββββββββββββββββββββββββββββ β
β β² β² β
β β (Ingests Contextual Signal Feeds) β β
β βββββββββββββββββββββββββββββββββ β β
β βΌ βΌ β β
β βββββββββββββββββββββββββββββββ βββββββββββββββββββββββββββββ β (Control Channel Commands) β
β β Threat Intelligence (CTI) β β Identity Governance / PKI β β β
β β EDR Posture / SIEM Alerts β β Continuous Risk Engine β β β
β βββββββββββββββββββββββββββββββ βββββββββββββββββββββββββββββ β β
ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββΌβββββββββββββββββββββββββββββββ
β
ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββͺβββββββββββββββββββββββββββββββ
DATA PLANE: POLICY ENFORCEMENT POINT (PEP) β
βΌ
ββββββββββββββββββββββββββββββββ ββββββββββββββββββββββββββββββββββββ ββββββββββββββββββββββββββββββββ
β UNTRUSTED SUBJECT β β POLICY ENFORCEMENT POINT (PEP) β β PROTECTED ENTERPRISE RESOURCEβ
β User / Workstation / Service βββββββββββββββΊβ API Gateway / Envoy Proxy / WAF βββββββββββββββΊβ App Workload / Relational DB β
β (Presents mTLS + Context) β β (Terminates connection, checks β β (Microsegmented Workload) β
β β β token, enforces PDP decision) β β β
ββββββββββββββββββββββββββββββββ ββββββββββββββββββββββββββββββββββββ ββββββββββββββββββββββββββββββββ
THE SEVEN TENETS OF ZERO TRUST ARCHITECTURE (NIST SP 800-207):
ββββββββββββββββ¬βββββββββββββββββββββββββββββββββββββββββββ¬ββββββββββββββββββββββββββββββββββββββ
β TENET NUMBER β AXIOMATIC TENET DEFINITION β CONCRETE ENTERPRISE IMPLEMENTATION β
ββββββββββββββββΌβββββββββββββββββββββββββββββββββββββββββββΌββββββββββββββββββββββββββββββββββββββ€
β Tenet 1 β All data sources and computing services β Treat local printer, cloud VM, and β
β β are considered resources. β laptop as individual secured assets.β
ββββββββββββββββΌβββββββββββββββββββββββββββββββββββββββββββΌββββββββββββββββββββββββββββββββββββββ€
β Tenet 2 β All communication is secured regardless β Mandatory Mutual TLS (TLS 1.3 mTLS) β
β β of network location. β inside internal Kubernetes clusters.β
ββββββββββββββββΌβββββββββββββββββββββββββββββββββββββββββββΌββββββββββββββββββββββββββββββββββββββ€
β Tenet 3 β Access to individual enterprise resourcesβ Ephemeral, per-session short-lived β
β β is granted on a per-session basis. β JWTs (5-minute expiry) via OAuth2. β
ββββββββββββββββΌβββββββββββββββββββββββββββββββββββββββββββΌββββββββββββββββββββββββββββββββββββββ€
β Tenet 4 β Access to resources is determined by β Attribute-Based Access Control β
β β dynamic policy (user, device, state). β (ABAC) evaluating device health & IPβ
ββββββββββββββββΌβββββββββββββββββββββββββββββββββββββββββββΌββββββββββββββββββββββββββββββββββββββ€
β Tenet 5 β The enterprise monitors and measures the β CrowdStrike/Wazuh EDR agent queries β
β β integrity and security posture of assets.β disk encryption, OS patch level. β
ββββββββββββββββΌβββββββββββββββββββββββββββββββββββββββββββΌββββββββββββββββββββββββββββββββββββββ€
β Tenet 6 β All resource authentication and authori- β Multi-Factor Authentication (FIDO2) β
β β zation are dynamic and strictly enforced.β re-prompted upon anomaly detection. β
ββββββββββββββββΌβββββββββββββββββββββββββββββββββββββββββββΌββββββββββββββββββββββββββββββββββββββ€
β Tenet 7 β The enterprise collects as much info as β Comprehensive SIEM log ingestion β
β β possible about assets to improve posture.β correlating Zeek, Sysmon & NetFlow. β
ββββββββββββββββ΄βββββββββββββββββββββββββββββββββββββββββββ΄ββββββββββββββββββββββββββββββββββββββ
ALLOW, CHALLENGE_MFA, or DENY.
from dataclasses import dataclass
from enum import Enum
from typing import Dict, Any
class AccessVerdict(Enum):
ALLOW = "ACCESS_GRANTED"
CHALLENGE_MFA = "STEP_UP_AUTHENTICATION_REQUIRED"
DENY = "ACCESS_DENIED"
@dataclass
class SubjectContext:
user_id: str
role: str
device_health_score: float # 0.0 to 1.0 (EDR checked: OS patched, BitLocker on)
is_corporate_managed: bool
source_ip_threat_score: float # 0.0 (Clean) to 1.0 (Known botnet / Tor exit node)
location_country: str
access_hour_utc: int
class ZeroTrustPolicyDecisionPoint:
def __init__(self, target_resource_criticality: str):
self.criticality = target_resource_criticality # "LOW", "MEDIUM", "HIGH", "CRITICAL"
def evaluate_access(self, context: SubjectContext) -> Dict[str, Any]:
"""
Dynamically evaluates SubjectContext against NIST SP 800-207 Tenets 4 & 6.
Computes composite Trust Score: T in [0.0, 100.0]
"""
# Base confidence calculation
trust_score = 0.0
# 1. Device Posture Evaluation (Weight: 40 points)
if context.is_corporate_managed:
trust_score += 20.0
trust_score += context.device_health_score * 20.0
# 2. Network & Threat Intelligence Scoring (Weight: 30 points)
threat_penalty = context.source_ip_threat_score * 30.0
trust_score += max(0.0, 30.0 - threat_penalty)
# 3. Behavioral Anomaly Signals (Weight: 30 points)
# Expected work hours: 04:00 UTC to 19:00 UTC
if 4 <= context.access_hour_utc <= 19:
trust_score += 15.0
# Expected geo-fencing: Domestic region
if context.location_country in ["IN", "SG"]:
trust_score += 15.0
# Dynamic Threshold Evaluation based on Resource Criticality
thresholds = {
"LOW": 40.0,
"MEDIUM": 60.0,
"HIGH": 80.0,
"CRITICAL": 92.0
}
required_score = thresholds.get(self.criticality, 80.0)
# Verdict resolution
if trust_score >= required_score:
verdict = AccessVerdict.ALLOW
elif trust_score >= (required_score - 20.0):
verdict = AccessVerdict.CHALLENGE_MFA
else:
verdict = AccessVerdict.DENY
return {
"user": context.user_id,
"resource_criticality": self.criticality,
"computed_trust_score": round(trust_score, 2),
"required_threshold": required_score,
"decision": verdict.value,
"signals": {
"device_posture_ok": context.device_health_score > 0.7,
"ip_clean": context.source_ip_threat_score < 0.2,
"geo_fenced": context.location_country in ["IN", "SG"]
}
}
# --- Academic Execution Demonstration ---
if __name__ == "__main__":
pdp_engine = ZeroTrustPolicyDecisionPoint(target_resource_criticality="CRITICAL")
# Scenario A: Legitimate engineer on compliant corporate machine
engineer_context = SubjectContext(
user_id="nikita.sharma@nikita-infotech.com",
role="Lead_DevSecOps",
device_health_score=0.95,
is_corporate_managed=True,
source_ip_threat_score=0.02,
location_country="IN",
access_hour_utc=10
)
decision_a = pdp_engine.evaluate_access(engineer_context)
print(f"[*] Scenario A: {decision_a['user']}")
print(f" Trust Score: {decision_a['computed_trust_score']}/100 β Verdict: {decision_a['decision']}")
# Scenario B: Stolen valid password, but executed via Tor from rogue device
adversary_context = SubjectContext(
user_id="nikita.sharma@nikita-infotech.com",
role="Lead_DevSecOps",
device_health_score=0.10,
is_corporate_managed=False,
source_ip_threat_score=0.90, # Known Tor node
location_country="RO",
access_hour_utc=2
)
decision_b = pdp_engine.evaluate_access(adversary_context)
print(f"[*] Scenario B (Adversary with valid password):")
print(f" Trust Score: {decision_b['computed_trust_score']}/100 β Verdict: {decision_b['decision']}")
# ==============================================================================
# ZERO TRUST KUBERNETES MICROSEGMENTATION: STRICT DEFAULT DENY + SELECTIVE ALLOW
# Enforces complete east-west workload isolation inside cluster namespaces
# ==============================================================================
# 1. BASELINE: DROP ALL INGRESS & EGRESS BY DEFAULT IN "PRODUCTION" NAMESPACE
apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
name: default-deny-all
namespace: production
spec:
podSelector: {} # Applies to every pod in the namespace
policyTypes:
- Ingress
- Egress
---
# 2. SELECTIVE ALLOW: DATABASE ACCESSIBLE ONLY FROM AUTHORIZED API WORKLOADS
apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
name: secure-postgres-microsegmentation
namespace: production
spec:
podSelector:
matchLabels:
app: relational-database
tier: storage
policyTypes:
- Ingress
ingress:
- from:
# Explicit pod selector: Only authenticated API service pods can connect
- podSelector:
matchLabels:
app: student-enrollment-api
security-tier: authorized-client
ports:
- protocol: TCP
port: 5432 # Standard PostgreSQL port
---
# 3. EGRESS LOCKDOWN: API SERVICE PODS CAN ONLY REACH DATABASE & DNS
apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
name: api-service-egress-lockdown
namespace: production
spec:
podSelector:
matchLabels:
app: student-enrollment-api
policyTypes:
- Egress
egress:
# Allow internal CoreDNS resolution
- to:
- namespaceSelector: {}
podSelector:
matchLabels:
k8s-app: kube-dns
ports:
- protocol: UDP
port: 53
# Allow traffic exclusively to PostgreSQL storage tier pods
- to:
- podSelector:
matchLabels:
app: relational-database
tier: storage
ports:
- protocol: TCP
port: 5432
import time
import hmac
import hashlib
import json
class SinglePacketAuthorizationEngine:
"""
Implements Single Packet Authorization (SPA) protocol concept.
Client transmits a single encrypted, signed UDP packet before TCP connection.
Firewall default state: 100% DROP all inbound SYN packets (Host is 'Dark').
Upon valid SPA packet verification: Gateway opens temporary dynamic pinhole.
"""
def __init__(self, pre_shared_hmac_key: bytes):
self.secret_key = pre_shared_hmac_key
def generate_spa_packet(self, client_ip: str, target_port: int) -> bytes:
payload = {
"source_ip": client_ip,
"target_service_port": target_port,
"epoch_timestamp": int(time.time()),
"client_nonce": "NONCE-9941a7c"
}
serialized = json.dumps(payload, sort_keys=True).encode('utf-8')
signature = hmac.new(self.secret_key, serialized, hashlib.sha256).hexdigest()
packet_envelope = f"{serialized.decode('utf-8')}||{signature}"
return packet_envelope.encode('utf-8')
def verify_and_authorize_pinhole(self, raw_packet: bytes, gateway_current_time: int) -> bool:
try:
payload_str, signature = raw_packet.decode('utf-8').split("||")
serialized = payload_str.encode('utf-8')
expected_sig = hmac.new(self.secret_key, serialized, hashlib.sha256).hexdigest()
# Constant-time comparison prevents timing attacks
if not hmac.compare_digest(expected_sig, signature):
print("[!] SPA Error: Invalid cryptographic signature! Packet dropped.")
return False
data = json.loads(payload_str)
# Replay protection: Packet must be within 30-second window
if abs(gateway_current_time - data["epoch_timestamp"]) > 30:
print("[!] SPA Error: Replay packet rejected! Timestamp expired.")
return False
print(f"[β] SPA Accepted: Dynamic firewall pinhole opened for {data['source_ip']} β Port {data['target_service_port']} for 15s.")
return True
except Exception as e:
print(f"[!] SPA Processing Error: {e}")
return False
# --- Academic Execution Demonstration ---
if __name__ == "__main__":
hmac_key = b"EnterpriseMasterclassSharedSecret2026"
spa_engine = SinglePacketAuthorizationEngine(hmac_key)
client_address = "198.51.100.44"
desired_port = 22 # Protected SSH Port (Hidden from Nmap)
# 1. Client mints SPA authorization packet
spa_knock = spa_engine.generate_spa_packet(client_address, desired_port)
print(f"[*] Generated SPA Knock Packet: {spa_knock[:60]}...")
# 2. Gateway verifies packet and temporarily unblocks port
now = int(time.time())
is_authorized = spa_engine.verify_and_authorize_pinhole(spa_knock, now)
assert is_authorized, "Legitimate SPA packet was rejected!"
# 3. Adversary attempts replay attack 2 minutes later
expired_time = now + 120
is_replay_accepted = spa_engine.verify_and_authorize_pinhole(spa_knock, expired_time)
assert not is_replay_accepted, "Replay attack should have been dropped!"
CONTINUOUS ADAPTIVE RISK & TRUST ASSESSMENT (CARTA) WORKFLOW:
[INITIAL STATE: USER AUTHENTICATES]
β’ Student signs in with valid password + primary factor.
β’ Action: GET /v1/student/profile (Low-risk operation).
β’ Policy Engine Evaluation: Trust Score = 65.
β’ Result: ALLOWED (Access granted to profile view).
β
βΌ
[INTERMEDIATE ACTION: MUTATION ATTEMPTED]
β’ Student attempts sensitive action: POST /v1/student/bank-account-update
β’ Policy Engine Evaluation:
βββ Action Risk Rating: HIGH
βββ Required Trust Score: 85 (Baseline 65 insufficient!)
βββ Context Drift Check: IP unchanged, but transaction type requires cryptographic proof.
β’ Result: STEP-UP CHALLENGE ISSUED (HTTP 403 Forbidden with Challenge header).
β
βΌ
[STEP-UP VERIFICATION: HARDWARE SECURITY KEY]
β’ Client receives step-up challenge: WWW-Authenticate: FIDO2-WebAuthn-Required
β’ User touches hardware security key (YubiKey / Passkey biometric prompt).
β’ Cryptographic challenge-response returned to Policy Enforcement Point.
β
βΌ
[STATE ADJUSTMENT & COMPLETE MEDIATION]
β’ Policy Decision Point re-evaluates: Trust Score elevated to 92.
β’ Action executed: POST /v1/student/bank-account-update β HTTP 200 OK.
β’ Ephemeral elevated permission automatically expires after 120 seconds.
SPIFFE / SPIRE CRYPTOGRAPHIC WORKLOAD ATTESTATION ARCHITECTURE:
1. WORKLOAD IDENTITY ASSIGNMENT (SPIFFE ID URI SCHEME):
Format: spiffe://<trust-domain>/ns/<namespace>/sa/<service-account>
β’ Student API Service : spiffe://nikita-infotech.com/ns/production/sa/student-api-sa
β’ Billing API Service : spiffe://nikita-infotech.com/ns/production/sa/billing-api-sa
β’ Database Service : spiffe://nikita-infotech.com/ns/production/sa/postgres-db-sa
2. WORKLOAD ATTESTATION PIPELINE:
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β KUBERNETES KERNEL WORKLOAD RUNTIME β
β 1. Pod launches on Node. SPIRE Agent queries Linux kernel (cgroups, PID, Namespace, UID). β
β 2. SPIRE Agent matches kernel attributes against SPIRE Server registration entries. β
β 3. SPIRE Server issues ephemeral X.509 SVID (SPIFFE Verifiable Identity Document). β
β 4. SVID certificate validity: Exactly 60 minutes (Auto-rotated every 30 minutes). β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β
βΌ (mTLS Transport Handshake)
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β ENVOY SERVICE MESH SIDE-CAR PROXY (PEP ENFORCEMENT) β
β Client Envoy initiates TLS 1.3 handshake presenting its SVID certificate. β
β Server Envoy validates SVID against trust-domain Root CA. β
β Authorization Policy Check: β
β IF client_spiffe_id == "spiffe://nikita-infotech.com/ns/production/sa/student-api-sa": β
β ALLOW TCP connection to Postgres port 5432 β
β ELSE: β
β TERMINATE TLS HANDSHAKE WITH ACCESS DENIED (Zero IP-based rules used!) β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
LEGACY VPN VS. ZERO TRUST NETWORK ACCESS (ZTNA) COMPARISON MATRIX:
βββββββββββββββββββββββββββ¬βββββββββββββββββββββββββββββββ¬βββββββββββββββββββββββββββββββ
β ARCHITECTURAL VECTOR β LEGACY CORPORATE VPN β ZERO TRUST NETWORK ACCESS β
βββββββββββββββββββββββββββΌβββββββββββββββββββββββββββββββΌβββββββββββββββββββββββββββββββ€
β Network Level Access β Layer 3 (Full Network Subnet)β Layer 7 (Application-Specificβ
β β Grants access to whole CIDR β Reverse proxy tunnel only) β
βββββββββββββββββββββββββββΌβββββββββββββββββββββββββββββββΌβββββββββββββββββββββββββββββββ€
β Trust Boundary β Extends edge perimeter to β Perimeter eliminated; trust β
β β the remote worker's home β evaluated per application β
βββββββββββββββββββββββββββΌβββββββββββββββββββββββββββββββΌβββββββββββββββββββββββββββββββ€
β Lateral Movement Risk β SEVERE: Attacker pivots from β ZERO: User cannot see or β
β β laptop to domain controller β ping other cluster IP nodes β
βββββββββββββββββββββββββββΌβββββββββββββββββββββββββββββββΌβββββββββββββββββββββββββββββββ€
β Device Health Posture β Checked once at login, or β Continuous real-time EDR β
β β completely uninspected β telemetry checks per request β
βββββββββββββββββββββββββββΌβββββββββββββββββββββββββββββββΌβββββββββββββββββββββββββββββββ€
β Asset Visibility β High: Internal ports answer β Dark Cloud: Protected apps β
β β to ping and Nmap scans β invisible without auth knock β
βββββββββββββββββββββββββββΌβββββββββββββββββββββββββββββββΌβββββββββββββββββββββββββββββββ€
β Authorization Model β Coarse-grained IP/Port ACLs β Fine-grained ABAC + JWT β
β β static firewall rules β dynamic risk scoring β
βββββββββββββββββββββββββββ΄βββββββββββββββββββββββββββββββ΄βββββββββββββββββββββββββββββββ
ZERO TRUST ARCHITECTURAL PITFALLS & DEFECT REMEDIATION:
// ==============================================================================
// PITFALL 1: THE PDP LATENCY BOTTLENECK (SYNCHRONOUS CENTRALIZED CHECK CHURN)
// SMELL: Every microservice sends a synchronous remote REST call to a single central
// PDP server on every HTTP request.
// PROBLEM: Increases latency by 150ms per hop; central PDP crash takes down all apps!
// REFACTOR: Deploy local Policy Enforcement sidecars (e.g., Open Policy Agent - OPA)
// that evaluate policies in-memory locally and synchronize rules asynchronously!
// ==============================================================================
// ==============================================================================
// PITFALL 2: CIRCULAR AUTHORIZATION DEADLOCKS IN SERVICE MESHES
// SMELL: Service A requires verification from Service B to authenticate.
// Service B requires verification from Service A to authorize.
// PROBLEM: Complete system lockup during container boot or cache flush!
// REFACTOR: Strict acyclic dependency graphs; separate Identity Issuance (SPIRE)
// from runtime operational microservice communication.
// ==============================================================================
// ==============================================================================
// PITFALL 3: THE "ZERO TRUST IN NAME ONLY" (ZTINO) FALLBACK VULNERABILITY
// SMELL: try { enforceZeroTrustCheck(); } catch (Exception e) { allowAccess(); }
// PROBLEM: When the policy server times out under heavy load, the system falls
// back to open access, completely bypassing security controls!
// REFACTOR: Adhere to Saltzer-Schroeder Fail-Safe Defaults: Always default to DENY
// on network timeout, unexpected exception, or missing contextual attributes!
// ==============================================================================
ZERO TRUST ARCHITECTURE (NIST SP 800-207) SPECIFICATION MATRIX:
βββββββββββββββββββββββββββ¬βββββββββββββββββββββββββββββββ¬βββββββββββββββββββ¬ββββββββββββββββββββββββββββββββ
β COMPONENT / STANDARD β IMPLEMENTATION MECHANISM β ARCHITECTURE TIERβ PRIMARY SECURITY OBJECTIVE β
βββββββββββββββββββββββββββΌβββββββββββββββββββββββββββββββΌβββββββββββββββββββΌββββββββββββββββββββββββββββββββ€
β Policy Engine (PE) β NIST SP 800-207 Brain β Control Plane β Computes access decision β
β Policy Administrator(PA)β Issues session tokens / certsβ Control Plane β Commands Enforcement Points β
β Policy Enforce Point(PEPβ Envoy Proxy, API Gateway β Data Plane β Terminates & gates traffic β
β Microsegmentation β K8s NetworkPolicy / eBPF β Network Layer β Blocks lateral attacker drift β
β Software-Defined Perim. β Single Packet Auth (SPA) β Transport Layer β Renders infrastructure dark β
β Workload Identity β SPIFFE / SPIRE X.509 SVID β Application Layerβ Cryptographic service identityβ
β Mutual TLS (mTLS) β TLS 1.3 Dual-Cert Handshake β Transport Layer β Mutual auth + confidentiality β
β Dynamic Trust Scoring β Multi-Signal Risk Engine β Control Plane β Contextual access gating β
β Step-Up Authentication β FIDO2 WebAuthn / Passkeys β Identity Tier β Adaptive risk elevation β
β Least Privilege Access β Fine-Grained ABAC Policies β Access Control β Restricts blast radius β
βββββββββββββββββββββββββββ΄βββββββββββββββββββββββββββββββ΄βββββββββββββββββββ΄ββββββββββββββββββββββββββββββββ€
β TOP 5 ZERO TRUST DESIGN AXIOMS: β
β 1. Assume Breach: Always architect under the premise that adversaries already reside on the network. β
β 2. Explicit Verification: Authenticate and authorize every transaction, subject, and device every time. β
β 3. Decouple Control Plane from Data Plane: PDP determines rules; distributed PEPs enforce them. β
β 4. Eliminate Implicit Perimeter Trust: Network location must never be used as a proxy for trust. β
β 5. Microsegment Everything: Isolate workloads so that compromise of one container stops lateral pivot. β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ