Secure Network Architecture & Components

CISSP CISSP · Communication and Network Security

By Abu Anns · Last updated 20 November 2026Exam version: April 2024 outline (current until 15 April 2027)

The OSI model as a security map

Domain 4 is built on the OSI model, and the CISSP expects you to use it as a map for placing controls and understanding attacks. The seven layers, from bottom to top, are: Physical (cables, signals, hubs), Data Link (switches, MAC addresses, Ethernet), Network (routers, IP addressing), Transport (TCP and UDP, ports), Session (managing connections), Presentation (data formatting, encryption, compression), and Application (HTTP, DNS, SMTP, and the protocols users interact with). The TCP/IP model collapses this into four layers (Link, Internet, Transport, Application), and the exam uses both. The practical value is diagnostic: when a question describes an attack or a control, identifying its layer tells you what it can and cannot do. A packet-filtering firewall works at Layers 3 and 4; it cannot inspect application content. A web application firewall works at Layer 7; it understands HTTP but adds latency. ARP spoofing is a Layer 2 attack; IP spoofing is Layer 3; a SYN flood is Layer 4; SQL injection is Layer 7.

TCP/IP fundamentals the exam tests

TCP is connection-oriented and reliable: it establishes a connection with a three-way handshake (SYN, SYN-ACK, ACK), guarantees delivery with sequencing and acknowledgements, and provides flow control. UDP is connectionless and unreliable: it sends datagrams with no handshake, no guarantee of delivery, and no ordering — which makes it faster and suitable for real-time traffic such as DNS queries, VoIP and video streaming. The exam tests the handshake because it is the basis for several attacks: a SYN flood sends SYN packets without completing the handshake, exhausting the server's half-open connection table; SYN cookies are the standard countermeasure. TCP sequence prediction allows session hijacking if sequence numbers are predictable. Common ports worth memorising: FTP 21, SSH 22, Telnet 23, SMTP 25, DNS 53, HTTP 80, POP3 110, NTP 123, NetBIOS 137–139, IMAP 143, SNMP 161/162, LDAP 389, HTTPS 443, LDAPS 636, and RDP 3389. The exam also tests IPv4 versus IPv6: IPv6 uses 128-bit addresses (versus IPv4's 32-bit), has no broadcast (it uses multicast instead), includes IPsec support natively in its design, and uses SLAAC (stateless address autoconfiguration) and Neighbour Discovery Protocol (NDP), which replaces ARP and introduces its own spoofing risks.

Firewalls: types and placement

A firewall controls traffic between network zones based on policy, and the exam tests the generations of firewall technology. Packet-filtering firewalls (first generation) examine packet headers — source and destination IP addresses and ports — against rules; they are fast but cannot understand application context or connection state. Stateful inspection firewalls (second generation) track the state of connections and allow return traffic for established sessions; they understand the TCP handshake and can block packets that do not belong to a known connection. Application-layer firewalls / proxies (third generation) inspect traffic at Layer 7, understanding protocols such as HTTP and FTP, and can enforce content policy; a proxy firewall terminates the connection on behalf of the client, so the internal host never communicates directly with the external host. Next-generation firewalls (NGFW) combine stateful inspection with application awareness, intrusion prevention, and user identity. A web application firewall (WAF) is a specialised Layer 7 firewall that protects web applications against attacks such as SQL injection and cross-site scripting. Firewall rules should follow a default-deny posture: everything is blocked unless explicitly allowed. The exam also tests firewall placement: a DMZ (demilitarised zone) is a network segment between the untrusted internet and the trusted internal network, hosting public-facing services such as web and mail servers so that a compromise of a public server does not give direct access to internal systems.

Network segmentation and micro-segmentation

Segmentation divides a network into zones so that a breach in one zone does not spread freely to others. The classic implementation is VLANs (virtual LANs), which separate traffic at Layer 2, combined with firewalls or ACLs controlling traffic between segments. A three-tier architecture — presentation (web servers), application, and data (databases) — is the standard pattern: each tier sits in its own segment, and traffic flows only between adjacent tiers through controlled paths. Micro-segmentation takes this further, applying policy at the workload level (often in software-defined networks), so that even two servers in the same VLAN cannot communicate unless policy allows it. The exam's framing is risk reduction: segmentation limits the blast radius of a compromise and is a key control for protecting sensitive systems such as payment processing (PCI DSS requires it) and industrial control systems. Air-gapping — physically isolating a network with no connection to any other network — is the strongest form of segmentation, used for the most sensitive systems, but it is expensive and inconvenient, and the exam notes that air gaps can still be bridged by removable media and insider action.

Intrusion detection and prevention

Intrusion Detection Systems (IDS) monitor traffic and alert on suspicious activity; Intrusion Prevention Systems (IPS) can also block it. The distinction matters: an IDS is passive (it watches a copy of the traffic, usually via a SPAN port or tap), while an IPS is inline (traffic passes through it, so it can drop malicious packets — but a failure can also block legitimate traffic, so fail-open versus fail-closed behaviour matters). Detection methods are signature-based (matching known attack patterns — accurate for known attacks, useless for novel ones) and anomaly-based / behavioural (detecting deviations from a learned baseline — can catch novel attacks but produces more false positives). Host-based systems (HIDS/HIPS) run on individual machines and can see encrypted traffic and system-level events; network-based systems (NIDS/NIPS) see traffic across the network but cannot inspect encrypted payloads. The exam's key insight: no detection system is perfect, and tuning — reducing false positives so that analysts pay attention to real alerts — is as important as the technology itself.

VPNs and remote access

A Virtual Private Network (VPN) creates an encrypted tunnel over an untrusted network. The two main protocols are IPsec and SSL/TLS VPN. IPsec operates at Layer 3 and has two modes: transport mode (encrypts only the payload, used for host-to-host communication) and tunnel mode (encrypts the entire packet including the header, used for site-to-site and remote access VPNs). IPsec has two protocols: AH (Authentication Header), which provides authentication and integrity but not confidentiality, and ESP (Encapsulating Security Payload), which provides confidentiality plus authentication and integrity — ESP is the one used in practice. Key exchange for IPsec is handled by IKE (Internet Key Exchange). SSL/TLS VPNs operate at the application layer (or via a tunnel) and are typically easier to deploy for remote users because they work through a web browser and pass through firewalls on port 443. Split tunnelling — where a VPN user sends some traffic through the tunnel and some directly to the internet — is a security risk the exam tests: it can allow an attacker to reach the corporate network through the user's unprotected internet connection, so the secure policy is to disable it.

Wireless security

Wireless networks are inherently exposed because radio signals cross physical boundaries, so encryption and authentication are essential. The history matters for the exam: WEP (Wired Equivalent Privacy) is fundamentally broken and must never be used. WPA was an interim fix using TKIP. WPA2 uses AES-CCMP and is the baseline for secure wireless; WPA3 is the current standard, adding Simultaneous Authentication of Equals (SAE) to protect against offline dictionary attacks and providing forward secrecy. Enterprise wireless should use 802.1X authentication with a RADIUS server, giving each user individual credentials rather than a shared pre-shared key (PSK). Common wireless attacks include evil twin (a rogue access point impersonating a legitimate one), rogue access points (unauthorised APs connected to the corporate network), wardriving (scanning for open networks), and jamming (denial of service by flooding the radio spectrum). The exam's practical points: disable WPS (Wi-Fi Protected Setup, which has a well-known brute-force weakness), use WPA2 or WPA3 with 802.1X for enterprise, and treat open Wi-Fi as untrusted — use a VPN.

Exam mindset for network architecture

Network questions reward layer thinking: identify the OSI layer of the attack or control, and the answer usually follows. Match the firewall type to the inspection depth required, remember that segmentation limits blast radius, and know the VPN modes cold — IPsec tunnel versus transport mode is one of the most reliable exam questions in this domain.

Advertisement

Exam traps to watch for

Practice questions

Test yourself on this guide. Answers are checked instantly, with a full explanation for each question.

Q1.An organisation wants to protect its public web application against SQL injection and cross-site scripting attacks. Which control is MOST appropriate?

Report an error

Q2.Two branch offices need to connect their networks over the internet so that all traffic between them is encrypted, including the original IP headers. Which IPsec configuration should be used?

Report an error

Q3.A server is overwhelmed by a flood of SYN packets that never complete the TCP handshake, exhausting its connection table. Which countermeasure is MOST effective?

Report an error

Q4.An attacker compromises a public-facing web server in an organisation’s DMZ. The internal database servers are unaffected because the firewall between the DMZ and the internal network blocks all inbound connections from the DMZ. Which design principle does this illustrate?

Report an error

Q5.A security team needs to detect a novel attack for which no signature exists yet. Which detection method is MOST likely to identify it?

Report an error

Q6.A company is deploying wireless networking for 500 employees. Which configuration provides the STRONGEST security?

Report an error

Q7.Remote workers connect to the corporate network via VPN. The VPN is configured with split tunnelling enabled, so users can access the internet directly while connected. What is the PRIMARY security risk of this configuration?

Report an error

Q8.An IPS is deployed inline between the internet and the internal network. During a power failure, the IPS stops processing traffic. Which behaviour is MOST appropriate for a high-security environment?

Report an error