CIS-1355 · Security Engineering · Module 9

Infrastructure Security

Layer network defenses, enforce boundaries between security zones, and control remote access.

Harden the infrastructure you already use

Cisco · Switch security and port security

Select controls by what they can see and enforce

Firewalls enforce policy on traffic entering or leaving a host or network boundary. They can allow, deny, reject, or log traffic depending on the policy and product. Check rule order, priority, and the default action. A stateless filter evaluates packets individually. A stateful firewall also tracks connection state, which can let it recognize return traffic for permitted connections. Host firewalls protect individual systems. Network firewalls can run as physical appliances, virtual appliances, or managed services. Both open-source and proprietary implementations exist.

WAF and Layer 7 firewall
A web application firewall inspects HTTP traffic to and from a web application. Other application-aware firewalls can identify or inspect additional protocols. Visibility into encrypted content depends on the deployment and where TLS terminates.
NGFW and UTM
A next-generation firewall typically adds capabilities such as application identification or intrusion prevention. Unified threat management packages several functions together, such as firewalling, IPS, web filtering, and sometimes email security or VPNs. Capabilities vary by product.
NAT gateway
Translates IP addresses and sometimes ports. A common cloud use is providing outbound connectivity from private workloads. Apply separate policy to control which traffic is allowed.
Forward and reverse proxies
A forward proxy handles outbound requests on behalf of clients. A reverse proxy receives requests on behalf of servers and routes them to backends. Either may apply security policy where its configuration and traffic visibility allow.
Network IDS and IPS
A network IDS alerts on suspicious traffic and can observe a copy from a network tap or switch mirror port. An inline network IPS sits in the traffic path and can drop matching traffic. Test blocking rules and plan for service impact if an inline device fails.
Deception and sinkholing
A honeypot is a monitored decoy. A honeynet is a group of decoy systems. High-interaction decoys expose realistic services and may include simulated activity or honeyfiles. A honeytoken is a decoy value or artifact whose use may trigger an alert. A DNS sinkhole can return a controlled address for a blocked domain, diverting traffic and helping identify affected clients. Isolate decoys and keep real credentials and sensitive data out of them.

Web filtering can evaluate URLs, categories, and reputation when the needed details are visible. DNS filtering can block resolution of particular names at a controlled resolver. File integrity monitoring (FIM) flags changes to selected files or configurations. Extended detection and response (XDR) correlates telemetry across sources such as endpoints, email, networks, and cloud services.

NIST · Firewalls and Firewall Policy (PDF) · NIST · Intrusion Detection and Prevention Systems (PDF)

Design and enforce security zones

A VLAN creates a logical Layer 2 broadcast domain that can span switches. Use a router ACL or firewall to enforce policy on routed traffic between VLANs or zones. A demilitarized zone (DMZ) places externally reachable services in a separate security zone. Explicitly allow the flows those services need to reach internal systems. Apply the same reasoning to other zones, such as management, guest, and database networks.

An air gap physically separates a system or network from other networks. Removable media and maintenance procedures still need controls. Zero trust grants access to specific resources based on explicit policy and signals such as identity and device posture. Network location alone does not determine trust. Reassess access as conditions change and apply least privilege to each resource.

NIST SP 800-207 · Zero Trust Architecture

Control remote access and device admission

A virtual private network (VPN) protects traffic routed through an encrypted tunnel across an untrusted network. With a full tunnel, all traffic covered by the routing policy goes through the VPN. With split tunneling, selected traffic goes through the VPN and other traffic uses a separate path. IPsec and TLS-based approaches are common. Require strong authentication and limit access to the resources the remote user needs. Traffic beyond the tunnel endpoints may need its own protection.

Network access control (NAC) can evaluate user or device identity and endpoint posture, such as patch status or endpoint protection. Depending on policy, the device can receive normal access, limited access, a remediation network, or a denial. Agent-based systems can collect posture from software on the device. Agentless methods can use other evidence, such as network profiling or information from existing management systems. Test admission policies with managed and unmanaged devices.

NIST · Guide to IPsec VPNs (PDF)

Practice case

A public portal and a remote administrator. An organization is deploying a public support portal. The web application needs to query one internal database. Administrators sometimes connect remotely, and an unmanaged laptop attempts to join the management network.

Where would you place the reverse proxy or WAF, the web application, and the database? Which paths would you allow? How should remote administrators connect, and what should happen to the unmanaged laptop?

Review a possible answer

Put the public reverse proxy or WAF in a DMZ and allow public HTTPS to it. Host the application and database in separate protected zones. Allow the proxy to reach the application on its configured port, and only the application hosts to reach the database on the required database port. Deny other cross-zone traffic by default. Use a restricted database account and log allowed application connections and denied attempts.

Require authenticated remote access with MFA and a managed administrator device. Limit the session to an approved management path, such as a jump host. Have NAC deny the unmanaged laptop access to the management network or place it in an isolated remediation zone. Test both permitted and blocked paths from the relevant zones.

Read and watch