Chapter 31: Network Security Architecture¶
Overview¶
Network security architecture defines how data flows safely through an organization — from the internet perimeter through internal segments to critical assets. The traditional castle-and-moat perimeter model has been supplanted by Zero Trust Architecture (ZTA), where no implicit trust exists based on network location. This chapter covers modern network segmentation, Zero Trust principles, next-generation firewalls, secure access service edge (SASE), intrusion detection and prevention, DNS security, and network traffic analysis for threat detection.
Learning Objectives¶
By the end of this chapter, students SHALL be able to:
- Design Zero Trust network architectures based on NIST SP 800-207 principles
- Configure and manage next-generation firewalls for micro-segmentation
- Implement secure remote access using ZTNA/VPN with strong authentication
- Deploy network-based threat detection using IDS/IPS and NTA
- Harden DNS infrastructure against DNS hijacking, poisoning, and tunneling
- Design SASE architecture for distributed enterprise networks
Prerequisites¶
- Strong networking fundamentals (TCP/IP, routing, VLANs, firewalls)
- Basic understanding of BGP and SD-WAN concepts
- Chapter 20 (Cloud Security) for cloud network security context
Why This Matters
The Colonial Pipeline ransomware attack was facilitated by a VPN account without MFA — a single credential compromise shut down 45% of East Coast fuel supply. The SolarWinds attack spread through trusted internal network connections because east-west traffic was largely unmonitored. Modern attackers exploit implicit network trust to move laterally after initial access. Zero Trust — verifying every connection, at every point, regardless of network location — is the fundamental architectural response to these threats.
31.1 Zero Trust Architecture¶
31.1.1 Zero Trust Principles (NIST SP 800-207)¶
graph TB
subgraph "Traditional Network Security"
T1[Trust inside\nthe perimeter]
T2[Perimeter\nFirewall]
T3[Implicit trust\nfor internal traffic]
end
subgraph "Zero Trust Architecture"
Z1[Never Trust\nAlways Verify]
Z2[Verify Explicitly\nIdentity + Device + Context]
Z3[Least Privilege\nMinimum access]
Z4[Assume Breach\nLimit blast radius]
end
subgraph "ZTA Components"
PE[Policy Engine\nAccess Decisions]
PA[Policy Administrator\nSession Management]
PEP[Policy Enforcement Point\nGateway / Proxy]
end
PE --> PA --> PEP
style Z1 fill:#e63946,color:#fff
style Z4 fill:#780000,color:#fff
style PE fill:#1d3557,color:#fff 31.1.2 Zero Trust Design Pillars¶
| Pillar | Traditional | Zero Trust |
|---|---|---|
| Identity | Username + password | MFA + identity risk score + device health |
| Device | Domain-joined = trusted | Every device must meet compliance policy continuously |
| Network | Inside = trusted | All traffic encrypted and inspected regardless of location |
| Application | IP-based access | Application-level micro-segmentation; per-app authorization |
| Data | Perimeter-protected | Data classification + DLP + encryption everywhere |
| Visibility | Perimeter logs | Complete telemetry from all resources and sessions |
31.1.3 NIST ZTA Reference Architecture¶
NIST SP 800-207 ZTA Implementation:
Resource Plane (what you're protecting):
├── Enterprise assets: servers, services, applications
├── Subject: users and devices requesting access
Control Plane (access decisions):
├── Policy Engine (PE): Brain — grant/deny/revoke access
├── Policy Administrator (PA): Communicates policies to PEP
├── Data sources: IdP, PKI, SIEM, CMDB, threat intel
Data Plane (traffic enforcement):
└── Policy Enforcement Point (PEP): Gateway that enforces decisions
Decision Flow:
Subject (user + device) → PEP → PA → PE → Access decision
PE checks: identity, device posture, behavioral analytics, threat intel
PE returns: Allow / Deny / Allow with conditions (MFA step-up)
PEP enforces decision; session monitored continuously
31.2 Network Segmentation¶
31.2.1 Segmentation Architecture¶
graph TB
subgraph "DMZ"
WEB[Web Servers\n443/80]
EMAIL[Email Gateway\n25/587]
DNS_EXT[External DNS]
end
subgraph "Internal Network — Segmented"
CORP[Corporate LAN\nUser Workstations]
SERVER[Server Segment\nInternal Apps]
DB[Database Segment\n1433/5432/3306]
IT[IT Management\nActive Directory, PKI]
PRIV[Privileged Access\nJump Servers, PAWs]
OT[OT/ICS\nAir-Gapped or DMZ-separated]
end
subgraph "Cloud Tenant"
VNET[Azure VNet / AWS VPC]
end
INTERNET --> FW_EXT[External Firewall\nNGFW]
FW_EXT --> DMZ
DMZ --> FW_INT[Internal Firewall\nNGFW]
FW_INT --> CORP
FW_INT --> SERVER
SERVER --> FW_DB[Micro-seg Firewall]
FW_DB --> DB
CORP <-..-> VPN_GW[ZTNA Gateway]
VPN_GW <-..-> VNET
style DB fill:#e63946,color:#fff
style PRIV fill:#780000,color:#fff
style OT fill:#f4a261,color:#000 31.2.2 Micro-Segmentation¶
Micro-segmentation extends traditional VLAN segmentation to per-workload policies:
# NSX-T / VMware micro-segmentation policy example
# Allow only specific traffic between application tiers
DFW Rule Set: 3-Tier App - Production
Rules:
- Name: Web-to-App
Source: Security Group "Web Tier"
Destination: Security Group "App Tier"
Service: HTTP (8080), HTTPS (8443)
Action: Allow
Direction: Inbound to App Tier
- Name: App-to-DB
Source: Security Group "App Tier"
Destination: Security Group "DB Tier"
Service: PostgreSQL (5432)
Action: Allow
- Name: Default Deny All
Source: Any
Destination: Any
Service: Any
Action: Deny
Log: True # Log all denied traffic for security analysis
31.3 Next-Generation Firewalls¶
31.3.1 NGFW Capabilities¶
| Feature | Traditional FW | NGFW |
|---|---|---|
| Filtering | IP/port/protocol | Application identity, user identity, content |
| SSL inspection | No | Yes (TLS termination + inspection) |
| IPS | Separate device | Integrated inline |
| URL filtering | No | Yes (75+ categories) |
| DNS security | No | Integrated |
| Cloud management | No | Yes (Panorama, Panorama Cloud, Strata) |
| Threat intelligence | No | Integrated (WildFire, Cortex XSOAR) |
| User identity | No | Integrated (AD, LDAP, SAML) |
31.3.2 Palo Alto Networks Policy Best Practices¶
NGFW POLICY DESIGN PRINCIPLES:
1. Application-based rules (not port-based)
WRONG: Allow TCP/80 from any to any
RIGHT: Allow "web-browsing" and "ssl" from Users to Internet
2. User-based policy
Allow "finance-team" to access "banking-sites" application
Block "guest-wifi" from internal resources
3. SSL/TLS inspection
Decrypt outbound HTTPS for inspection (with proper cert trust)
Exclude: financial sites (banking), healthcare (HIPAA concerns)
4. Default deny posture
Last rule: Block All from Any to Any with logging enabled
Log all blocked traffic to SIEM
5. Threat prevention profiles
Attach threat prevention profile to all allow rules
Anti-Spyware: Block critical/high, alert on medium
Vulnerability Protection: Block critical/high
Antivirus: Block on all alert threshold
Wildfire: Block on malicious
6. Application groups
Create custom application groups for:
- Approved collaboration: Teams, Zoom, Slack
- Development: GitHub, GitLab, npm
- Marketing: approved social media
31.4 Secure Remote Access¶
31.4.1 Traditional VPN vs. ZTNA¶
| Dimension | Traditional VPN | ZTNA |
|---|---|---|
| Trust model | Trusted once connected | Zero trust — verify per connection |
| Access scope | Full network access | Application-specific access only |
| MFA enforcement | Optional | Mandatory |
| Device posture | Not checked | Continuously evaluated |
| Lateral movement | Easy | Nearly impossible |
| Performance | Hair-pins through VPN | Direct-to-cloud or app-optimized |
| Examples | Cisco AnyConnect, OpenVPN | Zscaler ZPA, Cloudflare Access, BeyondCorp |
31.4.2 ZTNA Implementation (Cloudflare Access)¶
# Cloudflare Zero Trust Access Policy
Application: Internal HR Portal
Domain: hr.internal.example.com → tunneled to 10.0.1.50:443
Access Policy:
Name: HR Staff + IT Access
Action: Allow
Include (all must match):
- Identity Provider: AzureAD
- Group: HR-Department OR IT-Team
- Device Posture:
OS Version: Windows >= 10.0.19041
Disk Encryption: Required
Anti-Malware: Required (CrowdStrike running)
- Country: United States
Authentication:
- SAML SSO via Azure AD
- MFA: Conditional on risk score
Session Duration: 8 hours (re-auth required after)
Audit logging: All access logged to SIEM
31.5 Intrusion Detection and Prevention¶
31.5.1 IDS/IPS Placement¶
Optimal IDS/IPS placement:
1. External (perimeter): Between internet and DMZ
- Detects inbound attack patterns, scanning, known exploits
- High volume; tune carefully to avoid alert fatigue
2. Internal (east-west): Between network segments
- Detects lateral movement, C2, data exfiltration
- Most valuable for detecting advanced attackers
3. Critical asset protection: In front of database / financial servers
- Highly tuned ruleset for specific asset type
- Minimal false positives required
IPS Mode (inline — can block):
- Perimeter: Appropriate — block known malicious patterns
- Internal: Caution — false positive blocks can cause outages
IDS Mode (out-of-band — monitor only):
- Always safe — no impact on traffic
- Alert then investigate
31.5.2 Suricata Configuration¶
# Suricata — open source IDS/IPS
# /etc/suricata/suricata.yaml key settings
af-packet:
- interface: eth0
cluster-id: 99
cluster-type: cluster_flow
outputs:
- eve-log:
enabled: yes
filetype: regular
filename: /var/log/suricata/eve.json
types:
- alert:
payload: yes
payload-printable: yes
metadata: yes
http-body: yes
- http
- dns
- tls:
extended: yes
- files:
force-magic: yes
rule-files:
- /etc/suricata/rules/emerging-all.rules # ET Open rules
- /etc/suricata/rules/local.rules # Custom rules
# Detection rule example — detect Cobalt Strike beacon
alert http $HOME_NET any -> $EXTERNAL_NET any (
msg:"ET MALWARE CobaltStrike Beacon Pattern";
flow:established,to_server;
content:"Content-Type|3a 20|";
content:"application/octet-stream";
content:"Accept-Encoding|3a 20|identity";
content:!"Referer|3a 20|";
classtype:trojan-activity;
sid:2034999;
rev:1;
)
31.6 DNS Security¶
DNS is both the most important protocol on the internet and one of the most abused by threat actors.
31.6.1 DNS Attack Types¶
| Attack | Description | Defense |
|---|---|---|
| DNS Tunneling | Encode data in DNS queries/responses for C2 | DNS analytics, length/entropy monitoring |
| DNS Hijacking | Redirect queries to malicious resolver | DNSSEC, resolver monitoring |
| DNS Cache Poisoning | Inject false records into resolver cache | DNSSEC validation, randomize ports |
| DGA C2 | Malware uses randomly generated domains | Threat intel feeds, ML-based DGA detection |
| Lookalike Domains | Typosquatting in DNS | Brand monitoring, DMARC, typosquat alerts |
| Fast Flux | Rapidly rotate IPs for C2 | Reputation-based DNS filtering |
31.6.2 Protective DNS¶
# Cloudflare Gateway / Umbrella / Quad9 DNS filtering
# Change resolver to protective DNS:
# Cloudflare: 1.1.1.2 (malware blocking) / 1.1.1.3 (malware + adult)
# Quad9: 9.9.9.9 (threat intelligence blocking)
# Cisco Umbrella: Deploy as forwarder on internal resolvers
# DNSSEC validation
# Enable on internal resolvers:
# Unbound configuration:
cat >> /etc/unbound/unbound.conf << EOF
server:
module-config: "validator iterator"
auto-trust-anchor-file: "/etc/unbound/trusted-key.key"
val-log-level: 2
EOF
# Log all DNS queries for threat hunting
# Bind9 query log:
logging {
channel query_log {
file "/var/log/named/query.log" versions 5 size 20m;
print-time yes;
print-category yes;
severity info;
};
category queries { query_log; };
};
31.7 SASE Architecture¶
Secure Access Service Edge (SASE) converges wide-area networking and security into a cloud-delivered service.
graph TB
subgraph "Traditional WAN + Security Stack"
HQ[HQ] --> MPLS[MPLS\nExpensive Private WAN]
MPLS --> DC[Data Center\nFirewall, Proxy, IPS]
BRANCH[Branch] --> MPLS
REMOTE[Remote User] --> VPN[VPN]
VPN --> DC
end
subgraph "SASE Architecture"
HQ2[HQ] --> SASE[SASE Cloud\nZscaler / Netskope / Cloudflare One]
BRANCH2[Branch] --> SASE
REMOTE2[Remote User] --> SASE
CLOUD[SaaS/IaaS] --> SASE
SASE --> FW2[FWaaS]
SASE --> SWG[Secure Web Gateway]
SASE --> CASB2[CASB]
SASE --> ZTNA2[ZTNA]
SASE --> SD[SD-WAN]
end
style SASE fill:#1d3557,color:#fff 31.7.1 SASE Components¶
| Component | Function |
|---|---|
| SD-WAN | Dynamic path selection for branch connectivity |
| FWaaS (Firewall as a Service) | Cloud-delivered NGFW for all locations |
| SWG (Secure Web Gateway) | URL filtering, SSL inspection, DLP |
| CASB (Cloud Access Security Broker) | SaaS visibility, DLP, anomaly detection |
| ZTNA | Zero-trust application access |
| ZTDNS (DNS Security) | Protective DNS filtering |
31.8 Benchmark Controls¶
| Control ID | Title | Requirement |
|---|---|---|
| Nexus SecOps-NS-01 | Network Segmentation | Tiered segmentation; DMZ, servers, users, privileged access separated |
| Nexus SecOps-NS-02 | Zero Trust Remote Access | ZTNA or VPN with MFA + device posture checking for remote access |
| Nexus SecOps-NS-03 | East-West Monitoring | IDS/IPS coverage of internal network segments |
| Nexus SecOps-NS-04 | DNS Security | Protective DNS resolver; DNS query logging; DNSSEC where applicable |
| Nexus SecOps-NS-05 | NGFW Deployment | NGFW with threat prevention at all external boundaries |
| Nexus SecOps-NS-06 | Network Traffic Analysis | NTA solution capturing east-west traffic; 90-day retention |
Exam Prep & Certifications¶
Relevant Certifications
The topics in this chapter align with the following certifications:
- CCNP Security — Domains: Network Security Architecture, Firewalls, IDS/IPS
- GIAC GCIH — Domains: Incident Handling, Network Defense, Intrusion Detection
Key Terms¶
NGFW (Next-Generation Firewall) — A firewall that includes traditional packet filtering plus application identification, user identity awareness, SSL inspection, and integrated IPS.
SASE (Secure Access Service Edge) — A cloud-delivered security architecture converging SD-WAN with comprehensive security services (SWG, CASB, ZTNA, FWaaS) into a single platform.
Zero Trust Architecture (ZTA) — A security model based on the principle of "never trust, always verify" — requiring explicit verification of identity, device posture, and context for every access request.
ZTNA (Zero Trust Network Access) — Technology that provides application-specific, identity-verified remote access without placing users on the corporate network — replacing traditional VPN.
Micro-segmentation — Granular network segmentation at the workload or application level using software-defined policies, rather than relying solely on VLANs and physical network boundaries.