Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
A host-based firewall filters traffic at the device itself rather than at a network boundary, controlling which services are reachable and which processes may connect out, wherever the device happens to be.

What Is a Host-Based Firewall?
A host-based firewall runs on the device it protects and decides, packet by packet, what is allowed in and out. Unlike a network firewall, which sees traffic only as it crosses a boundary, it is present wherever the device is: in the office, on a hotel network, at home, or inside the same flat server subnet as four hundred other machines that a perimeter device will never arbitrate between.
Most organisations filter north-south traffic well and east-west traffic barely at all. Once an attacker has a foothold on one workstation, the interesting activity, which is SMB, RDP, WinRM, and WMI connections between peers on the same segment, never passes a firewall at all. A host firewall with a default-deny inbound policy is the control that turns that flat segment into something closer to segmentation without rewiring anything.
WannaCry in May 2017 illustrated the point at scale. The worm propagated over TCP 445 between machines that had no business speaking SMB to each other, and a workstation policy blocking inbound 445 from peer devices would have stopped the spread regardless of the patch state of the host.
Windows Defender Firewall with Advanced Security is considerably more capable than its control panel suggests. It supports three profiles, applied automatically by network type, so that a laptop enforces a stricter policy on an untrusted network than on the corporate domain. Rules can be scoped by remote address, by interface type, by service, and by application path, and connection security rules can require IPsec authentication so that only domain-joined machines may connect to a given port. It is managed through Group Policy or Intune and logs dropped packets to a file that is almost never collected but should be.
On Linux, nftables has replaced iptables as the kernel framework, with firewalld and ufw as the friendlier front ends; the INPUT chain is where the equivalent default-deny policy belongs. BSD systems use pf, and macOS has both an application-level firewall and the underlying packet filter. Cloud workloads add a second layer that behaves similarly: security groups and network security groups are host-attached filters in all but name, and the same default-deny discipline applies.
Cyber Essentials treats this explicitly. The firewalls control requires boundary firewalls and, for devices used on untrusted networks such as home or public Wi-Fi, a software firewall enabled and configured on the device itself. For most organisations with hybrid working, that means the host firewall is in scope for every laptop, not just for servers.
PCI DSS requirement 1 similarly expects controls on devices connecting from outside the network, and ISO 27001 Annex A 8.20 on network security is straightforward to evidence when host policy is centrally managed and reported. The NCSC device security guidance assumes a host firewall in its baseline builds for Windows and macOS.
The failure we find most often is not an absent firewall but an unmanaged one: the service enabled, the default profile untouched, and a long tail of rules added by application installers over several years. Our Infrastructure and Architecture Review Services review host firewall policy alongside segmentation design, because the two controls only make sense together.
Yes. A network firewall filters traffic crossing a boundary and sees nothing of the traffic between devices on the same segment, which is exactly the path used for lateral movement. A host firewall also travels with the laptop onto home and public networks, where no corporate boundary exists at all. The two controls cover different traffic and neither substitutes for the other.
It is worth doing, starting with servers. Restricting outbound connections to known destinations disrupts command and control channels and data exfiltration, and on a server the legitimate destination list is usually short and stable. On user devices the list is long and changes constantly, so begin with blocking direct outbound SMB and RDP to the internet, then expand gradually.
Usually remote support tools, file and printer sharing, and some monitoring agents that poll the device rather than reporting to a collector. Run the policy in a logging-only configuration on a pilot group first, review the dropped connections, and write scoped rules for the legitimate sources. Most estates find fewer genuine exceptions than expected, and most of them are administrative tooling.
Functionally it is close. A security group is a stateful filter attached to an instance or interface and enforced by the platform, so it gives you default-deny inbound without touching the guest operating system. Many teams rely on it alone, which is reasonable, though keeping the in-guest firewall configured as well gives defence in depth if a security group is later widened by mistake.
You May Be Interested In These
We use a few cookies that are needed to make this website work, and they are always on. We would also like to set analytics cookies, to see which pages are useful; functional cookies, so that our live chat works; and advertising cookies, to measure our campaigns. We will not set any of those until you say yes, and you can change your mind at any time from the Cookie Settings link in the footer.
Choose the cookies you are happy for us to use. Strictly necessary cookies are always on; everything else is off until you turn it on.
These cookies keep the website working: they hold your security checks, your form submissions and the choice you make here. The site cannot run without them, so they are always on.
These let our live chat open and remember your conversation, and let the newsletter sign-up work. Switch them off and the chat window will not appear.
These tell us which pages are read and how people find us, so that we can improve the site. The reports are about groups of visitors, never about you personally.
These let us measure our advertising and show you our adverts on other websites, such as LinkedIn, Google and Meta. Switch them off and you will still see adverts, but they will be less relevant.