Skip to content

DOMAIN 4 · Communication and Network Security

Network segmentation and firewall policy

Connect trust boundaries and allowed traffic flows to firewall rules, logging, testing, and controlled operational change.

Network segmentation divides environments so that communication can be limited according to purpose and risk. Begin with a flow inventory: source, destination, protocol, service, direction, owner, and business justification. A diagram without an explicit communication policy can conceal unrestricted paths.

NIST SP 800-41 Rev. 1 explains firewall technologies, policy, and deployment. It recommends denying traffic unless explicitly permitted by policy.1 The document dates from 2009; use it for these foundational concepts rather than treating its product examples as a current procurement guide.

A subnet or VLAN creates a logical division. It does not by itself prove that sensitive traffic is restricted. Routing rules, access-control lists, firewalls, host controls, and application permissions determine what is actually reachable.

Treat a trust boundary as a place where assumptions change. A guest wireless network, employee endpoint network, public service zone, and administration network may need different permissions. Give each a defined purpose and assess paths crossing between them, including management and monitoring traffic.

Function Information used Important limitation
Packet filtering Headers such as addresses, protocol, and port Does not necessarily understand application behavior
Stateful inspection Packet information plus connection state A permitted connection can still carry harmful application actions
Application proxy Application-level interaction through an intermediary Adds application-specific behavior and operational dependencies
Host firewall Rules close to an individual endpoint Configuration and coverage must be maintained on each relevant host

These functions may be combined in one implementation. NIST discusses their capabilities and tradeoffs; a product label alone does not establish which protections are enabled.1

A demilitarized zone, or DMZ, provides a separate zone for services exposed to less trusted networks. The important property is controlled communication, not simply the zone’s name. Restrict paths from public-facing services to internal systems and separate administration from ordinary service traffic.1

An Internet-facing web service should not receive broad access to an internal database network merely because it needs one database connection. Evaluate outbound traffic too: unrestricted egress may allow compromised hosts to contact unwanted destinations or send information out.

An illustrative policy permits clients to contact the front end on its approved service. The front end can contact a specific application service. Only that application service reaches the database on its required port. Administrative access comes from an approved management path with a separate identity and logging.

Record this as a small allowed-flow matrix before translating it into rules. Test both permitted flows and meaningful denied flows: a client reaching the database, a web host opening arbitrary internal connections, and an unauthorized host accessing management interfaces. This is a teaching architecture, not a universal production template.

Every exception should have an owner and purpose. Remove expired rules and investigate overly broad or shadowed rules. Test configuration changes, provide a rollback approach, and preserve appropriate logs. Firewall maintenance is an ongoing activity because hosts, services, and communication needs change.1

Also consider failure and bypass paths. Redundant firewalls provide little protection if an alternate network route skips enforcement. A blocked connection can be correct security behavior or an operational mistake; context and records are needed to tell them apart.

Network address translation is not a replacement for an access policy. Segmentation does not eliminate the need for application authorization or encryption. A rule allowing a port proves only that the specified traffic may pass; it does not prove the application is safe or the user is authorized.

See secure design for resource-level trust decisions and TLS for protecting permitted communication.

  1. NIST, SP 800-41 Rev. 1: Guidelines on Firewalls and Firewall Policy, sections 2–5. ↩ ↩2 ↩3 ↩4

Independent, unofficial study notes. Not endorsed or reviewed by ISC2. CISSP is a registered trademark of ISC2.

Editorial policy & attribution