// NETWORKS · FAQ · HANS STUDY
Network hardening questions, answered
These are the questions people ask most about hardening switches, access ports, management access, segmentation and Windows servers. Each answer is short, and the longer baselines and courses are linked where they exist.
Switches
What is network hardening?
Network hardening is removing what a device does not need and locking down what it does, so there is less to attack and less to go wrong. On a switch that means secure management access, guarded access ports, segmentation, and logs sent somewhere central. The same idea applies to firewalls and servers, and the controls below are the ones that come up most.
What should a switch hardening baseline include?
A baseline covers the management plane, the access ports, segmentation, time and logging, and a recovery path. Concretely that is SSH only, central authentication with a local fallback, SNMPv3, authenticated NTP, syslog off the box, port security, BPDU guard, DHCP snooping, ARP inspection, and unused ports shut down. The switch configuration audit checklist lists each item as a pass or fail check.
What are the Cisco switch security best practices?
Use SSH version 2 only, restrict the VTY lines to a management subnet, and use TACACS+ with a local fallback and a type 9 secret. Turn off the HTTP server and Smart Install, then add port security, BPDU guard, DHCP snooping and Dynamic ARP Inspection on access ports. The Catalyst IOS-XE hardening baseline has the full set of commands for a Catalyst 9200 or 9300.
How do I harden an Aruba CX switch?
Put SSH on the management VRF behind an allow-list, authenticate with TACACS+ and keep local as the last method, and set SNMP to v3 only. On access ports add port security, BPDU guard and loop protection, then enable DHCP snooping and ARP inspection per VLAN. Syntax differs by series and release, so use the AOS-CX hardening baseline and check it against the CLI guide for your model.
What should I do with unused switch ports?
Shut them down and move them to a blackhole VLAN with no routed interface. Shutting the port stops the link, and the blackhole VLAN limits the damage if someone later enables the port without checking. Add a description such as UNUSED so the port is easy to spot in an audit.
Access ports
What does port security do, and what are its limits?
Port security limits how many MAC addresses a port will accept and can shut the port when a different one appears. A maximum of 1 catches an unmanaged switch plugged in to add drops, because each device behind it presents its own source MAC. It is not authentication: MAC addresses can be spoofed, so it deters casual plug-ins but not a determined attacker.
What is BPDU guard, and does it replace port security?
BPDU guard shuts down an edge port that receives a spanning tree BPDU, which catches a managed switch or a loop where only a camera or workstation should be. It does not replace port security. The two cover different cases, so access ports normally get both.
How does DHCP snooping work?
DHCP snooping lets the switch accept DHCP server replies only on ports you mark as trusted, which are normally the uplinks. That blocks rogue DHCP servers, and it builds a binding table of IP address, MAC address and port that other features use. If you forget to trust the uplink, the VLAN loses its DHCP server, so make the change in a maintenance window.
What is Dynamic ARP Inspection, and why does it break static-IP devices?
Dynamic ARP Inspection drops ARP packets that do not match the DHCP snooping binding table. A camera or door controller with a static address never asked a DHCP server, so it has no binding and its ARP is dropped once DAI is on. Use DHCP reservations where you can, and add ARP ACLs or static bindings for devices that must stay static before you enable DAI on the VLAN.
Is 802.1X better than port security?
Yes, because 802.1X authenticates the device or user before the port passes traffic, while port security only counts MAC addresses. Port security is a reasonable stopgap where 802.1X is not deployed. If you run 802.1X, check your vendor guidance before stacking both on the same port.
How do I handle devices that cannot do 802.1X?
Use MAC Authentication Bypass where the switch and your policy server support it, and treat it as weaker than 802.1X. The switch falls back to the device MAC address, which can be spoofed. Limit the exposure with a dedicated VLAN for cameras or printers, restrictive ACLs, and a known-device list.
Management plane
Should I disable Telnet and HTTP on switches?
Yes. Telnet sends passwords in the clear and HTTP sends session cookies in the clear, and encrypted replacements exist. Use SSH, and keep the HTTPS interface only if you use it, restricted to the management network with a real certificate.
Should I use SNMPv3 instead of SNMPv2c?
Yes, if your monitoring tool supports it. SNMPv2c authenticates with a community string that crosses the network unencrypted, while SNMPv3 adds user-based authentication and encryption. Remove the default communities, set a v3 user with authentication and privacy, and confirm every monitoring system speaks v3 before you switch v2c off.
Do I need TACACS+ or RADIUS for switch logins?
Central authentication is worth having because it gives you named accounts, accounting and one place to remove a leaver. Always keep a local fallback account so a down AAA server does not lock you out, and test it from the console before you close your session. On Aruba CX, command authorization has no fail-through, so prove the setup on a lab switch first.
Why authenticate NTP and send syslog off the switch?
Logs are only useful if the timestamps can be trusted and the logs survive the device. Authenticated NTP stops a spoofed time source from skewing your evidence, and a syslog collector off the switch keeps the record when a local buffer rolls over or the switch is tampered with. Pin the syslog source to a management interface so searches by switch address keep working.
How should I handle switch firmware updates?
Check the target release against the vendor security advisories before you deploy, then upgrade in a way that can roll back. On Catalyst that means install mode, and on AOS-CX it means staging the new image on the secondary partition and keeping the known-good one. Back up the configuration first.
Segmentation
How should I segment cameras and IoT devices?
Give each function its own VLAN: management, systems, cameras, access control and an unused-port blackhole. Cameras should talk only to the VMS or recorder and should have no internet access, since a camera that can reach the internet is a path out for an attacker. The VLAN segmentation guide shows the layout and the firewall rules between zones.
Is VLAN segmentation enough on its own?
No. A VLAN separates broadcast domains, but the policy comes from whatever routes or filters between them, such as a firewall or ACLs. Without rules between VLANs, a routed network is only nominally segmented. Also disable dynamic trunking and keep the default VLAN unused.
Windows servers and audit policy
How do I harden Windows Server 2016, 2019 or 2022?
Start from a published baseline, such as the Microsoft Security Compliance Toolkit, DISA STIG or CIS Benchmark settings, and apply it through Group Policy to a test OU first. Then cover accounts and privileges, the attack surface, protection features, and logging. The Group Policy baseline walks through it, and the Server 2025 changes are covered separately.
What are the CIS Benchmarks, and which profile should I use?
CIS Benchmarks are consensus configuration guides published by the Center for Internet Security, with separate Windows Server benchmarks per version. They define Level 1 and Level 2 profiles, with separate profiles for member servers and domain controllers. Level 1 is the practical starting point, and Level 2 adds settings that can affect functionality, so test it before rollout.
What is the difference between Audit logon events and Audit account logon events?
Audit logon events records logons on the computer that was accessed, while Audit account logon events records credential validation by the authority that checked the account, which is a domain controller for domain accounts. These are the legacy basic settings. In Advanced Audit Policy the equivalents are the Logon, Logoff and Special Logon subcategories under Logon/Logoff, and Credential Validation and the Kerberos subcategories under Account Logon.
Should I use basic or Advanced Audit Policy?
Use Advanced Audit Policy Configuration and leave the basic categories alone. It has finer subcategories, and Microsoft says basic and advanced settings should not be mixed. If you apply advanced settings through Group Policy, enable the setting "Audit: Force audit policy subcategory settings (Windows Vista or later) to override audit policy category settings".
Which audit settings give the most value on a server?
Logon, account management, policy change, special logon and process creation are the core set. Process creation produces event 4688, and it is far more useful when you also enable "Include command line in process creation events". Logon produces 4624 and 4625, special logon produces 4672, and a change to audit policy itself produces 4719. The audit logging guide has the full auditpol baseline.
How do I audit changes to Group Policy objects?
Enable Audit Directory Service Changes on the domain controllers and make sure the GPO objects have the auditing entries you need, then watch event 5136. A change to the audit policy itself shows up as event 4719 under Audit Policy Change. For anything close to real-time, forward those events to a central collector instead of reading them on each domain controller.
Do Sysmon logs overlap with Windows audit logs?
Partly. Both can record process creation, but Sysmon adds file hashes, network connections and other detail that native auditing does not, while Windows auditing covers logon, account and policy events that Sysmon does not. Most teams run both, and forward them off the server so a compromised host cannot erase the record.
Reviews and audits
How often should firewall rules be reviewed?
At least twice a year, and again after any significant change. PCI DSS 4.0 requires a review of network security control configurations at least every six months (requirement 1.2.7), and that is a sensible cadence for most organizations. Record who reviewed, what changed and who signed off.
What does a firewall rule review check?
It checks each rule against its stated purpose: unused and shadowed rules, any-any entries, NAT and port-forward exposure, logging, administrative access, firmware status, and segmentation between zones. Every rule should have an owner and a business reason. The firewall review covers this as a rule-by-rule written report.
How do I audit a switch configuration?
Compare the running configuration against a written baseline, setting by setting, and record each gap with a fix. Check management access, SNMP, logging and time first, then access-port protections, VLAN use and unused ports. The free switch config tool helps with base configurations, and the switch configuration audit is a remote review across a whole estate.
What is the difference between a network assessment and a firewall or switch review?
A firewall or switch review looks at one device type in depth, while a network security assessment compares segmentation, the rule base, switch configurations, remote access, MFA and logging against the standard you owe. Use a device review for a known concern and an assessment for a gap list across the whole network.
How should I harden the servers behind a video management system?
Treat them like any other critical server: apply a baseline through Group Policy, restrict access, enable Advanced Audit Policy, patch on a schedule, and keep them in their own systems VLAN. The VMS server hardening service applies these controls to VMS hosts.
Go deeper
Practise the controls in the free Network Hardening Level 1 and Windows Hardening Level 1 courses, or take the Network and Windows Hardening training.
Want a second pair of eyes on your switches or servers? Book a working session.