ExplainerProtective DNS
Blocking malware domains before the blocklists catch up
You will be able to decide where protective DNS fits in your security stack and how to close the gap before blocklists list new malicious domains. You can set per-network and per-VLAN policy, plan for encrypted DNS bypass and cached answers, and verify that blocking works on every segment.
Watch the explainer
Blocking malware command and control at the DNS layer is refusing to resolve domains malware depends on. No answer means no connection, so downloads and callbacks fail before they start. Protective DNS, the model behind federal protective DNS services, applies this at the resolver. Purple data shows AI-driven detection flags threats up to 10 days before standard blocklists.
What is command-and-control blocking at the DNS layer?
Command and control (C2) is the channel a compromised device uses to receive instructions and return stolen data. MITRE ATT&CK catalogs it as a distinct tactic, TA0011. Most malware finds its C2 server by name, not by fixed IP address. Names let attackers move infrastructure without rebuilding the malware.
Every name must be resolved before a device can use it. The Domain Name System, defined in RFC 1034 and RFC 1035, turns a domain into an address. Your devices send that question to a recursive resolver, usually the one your network hands out through DHCP.
Protective DNS turns that resolver into a security control. It checks each lookup against threat intelligence and policy, then refuses to answer for malicious domains. In the US, CISA offers a protective DNS resolver to federal civilian agencies.
Three kinds of domain matter most for protective DNS malware blocking:
- C2 callbacks. The domain an infected device contacts for orders, updates or a place to send data.
- Drive-by download hosts. Sites that push a malicious payload to a visiting browser, or that a phishing link points at.
- Newly registered malicious domains. Fresh names created for one campaign, used briefly and then abandoned.
Placement is the point. A firewall judges a connection to an IP address the device already knows. Protective DNS acts one step earlier, when the device asks where to go. If no answer arrives, there is no connection for the firewall to judge and no payload for antivirus to scan.
How does DNS filtering block malware before a connection opens?
The resolution path, step by step
- A device on your guest or staff WiFi needs a domain. The request might come from a browser tab, an app or malware calling home.
- The query goes to the resolver your network assigns. With protective DNS in place, that resolver applies your policy.
- The resolver checks the domain against threat intelligence, its classification and the policy for that network or VLAN (virtual LAN, a logically separated segment of one physical network).
- A clean domain resolves normally, through recursive lookups to its authoritative servers.
- A malicious or suspect domain gets a blocked answer instead of a usable address. The resolver logs the query with the network it came from.
- The malware has nowhere to connect. The callback fails, the download never starts, and the event appears in your reports.
A device asks the protective resolver for a domain; malicious domains get a blocked answer and no connection opens, while clean domains resolve and connect normally.
Connections
- Device on guest or staff WiFi to Protective DNS resolver (DNS query)
- Protective DNS resolver to Policy and threat check (evaluate)
- Policy and threat check to Blocked answer (malicious or suspect)
- Blocked answer to No connection opens (device has no address)
- Policy and threat check to Normal resolution (clean)
- Normal resolution to Connection proceeds (IP address returned)
Command and control blocking also gives you a signal. A device that keeps asking for a blocked C2 domain is probably infected. The resolver log points you to the network and the time, before its callback succeeds.
Why do malware blocklists lag behind new domains?
Classic blocklists are signature tools. They list a domain once someone has seen it cause harm. That takes a chain of events. A campaign has to run, and someone has to capture a sample or a report. An analyst or automated system then confirms it, the feed publishes, and every resolver pulls the update.
Attackers design for that gap. Domain generation algorithms (DGAs) produce long lists of candidate names, and the malware tries them until one resolves. MITRE catalogs this as T1568.002. Fast flux rotates the IP addresses behind a name to frustrate address-based blocking. Phishing and dropper campaigns often register a fresh domain, use it briefly, then move on.
The first hours and days are when a new malicious domain does its damage. That is also when a list-only tool has nothing to match against. An unlisted malicious domain looks exactly like a legitimate new site.
A malicious domain is registered, goes live and is resolved by devices; behavioral detection blocks it during the gap, while the signature route waits for analysis and publication.
Connections
- Domain registered to Campaign goes live (weaponized)
- Campaign goes live to First devices resolve it (lookups begin)
- First devices resolve it to Behavioral detection flags it (AI-driven route)
- Behavioral detection flags it to Blocked inside the gap (block at first lookups)
- First devices resolve it to Sample captured and analyzed (signature route)
- Sample captured and analyzed to Listed on standard blocklists (review and publication)
How behavioral detection closes the gap
Closing the gap means judging a domain on how it looks and behaves, not only on whether anyone has reported it. Protective DNS services commonly score signals such as these:
- Domain age and registration pattern. Few legitimate sites are hours old when your staff first visit them.
- Lexical structure. DGA output tends to look random to a model trained on real domain names.
- Hosting and infrastructure. Shared addresses or name servers with names already tied to abuse raise the score.
- Query behavior. Many devices suddenly asking for a name nobody requested yesterday is a warning sign.
Shield applies AI-driven detection at resolution time. Purple data shows it flags threats up to 10 days before they reach standard blocklists. Those 10 days sit inside the gap where list-only tools still allow the domain. This is zero-day domain blocking: stopping a domain on first sight, before any feed has a record of it. Detection is probabilistic, not certain, which is why policy and verification matter.
Blocking newly registered domains as a policy
Some teams add a blunter rule: block any domain registered within a set number of days. Newly registered domain blocking catches campaigns that behavioral scores miss. The cost is occasional friction with legitimate new sites. It suits staff and device networks, where browsing is predictable, better than open guest networks. Our guide to newly registered domain blocking covers how to set the age threshold.
Protective DNS, DNS firewall or blocklist: which terms get confused?
Vendors use these terms loosely, and buyers inherit the confusion. The useful distinction is between a list and a decision. A blocklist records what someone has already seen. A protective resolver decides, at the moment of the lookup, whether a domain should resolve.
| Term | What it is | Where it acts | What it catches | What it misses |
|---|---|---|---|---|
| Protective DNS | A recursive resolver that applies security policy and threat detection to every lookup | At resolution, before any connection | C2 domains, malware hosts, phishing and suspect new domains | Hard-coded IPs and threats already on the device |
| DNS filtering | Policy on which categories of domain resolve, often for content control | At resolution | Unwanted categories such as adult or gambling sites | New threats, unless paired with threat detection |
| DNS firewall (RPZ) | A resolver feature that rewrites answers using Response Policy Zones | At resolution, on a resolver you operate | Domains on the feeds you load | Anything newer than your last feed update |
| Threat blocklist feed | A published list of known malicious domains or IPs | Nowhere on its own; it feeds other controls | Confirmed, previously seen threats | Domains in the gap before listing |
| Network firewall | A device or service that permits or denies traffic flows | At connection, on IPs, ports and sessions | Known malicious IPs and disallowed ports | Fresh domains on shared or rotating IPs |
| Endpoint protection | Antivirus or EDR software running on the device | On the managed device, after arrival | Malicious files, processes and behavior | Unmanaged guest devices |
A DNS firewall built on Response Policy Zones, described in the IETF draft draft-vixie-dnsop-dns-rpz, is only as current as its feeds. Protective DNS adds detection on top of those feeds. Our guide to protective DNS vs DNS filtering vs DNS firewall covers the vocabulary in full.
Where do you meet malware domain blocking in practice?
Guest WiFi you do not manage
Hotel guests, retail shoppers and airport passengers bring devices you will never manage. You cannot install antivirus on them or audit their patch levels. DNS is the one control you apply to every device on the network, because every device has to resolve names. Blocking malware domains here protects your brand and stops infected devices using your connection to call home. Whitbread and AGS Airports run Shield on their networks.
Healthcare and the public sector
Public sector organizations were early adopters, partly because CISA provides Protective DNS services. Healthcare networks mix clinical systems, staff laptops, and patient and visitor devices on shared infrastructure. Per-network policy lets you hold clinical segments to a strict rule set while guest segments stay usable.
Staff networks and device segments
On managed laptops, endpoint protection is already in place, and DNS blocking stops threats one step earlier. Device segments are a different case. Cameras, digital signage, and payment terminals often cannot run endpoint software at all. A tight policy per VLAN restricts what those devices can resolve. If card terminals sit in scope for PCI-DSS, narrowing their reachable domains reduces the exposure of that segment.
What are the limits of DNS malware blocking?
Traffic that never asks DNS
Malware that connects to a hard-coded IP address skips name resolution entirely. Some families use peer-to-peer networks for C2. Protective DNS sees neither. Address-based firewall rules and endpoint protection cover this ground.
Encrypted DNS that bypasses your resolver
DNS over HTTPS (RFC 8484) and DNS over TLS (RFC 7858) encrypt lookups. Both protect privacy on hostile networks. When a browser or app sends them to its own resolver, they also skip your policy.
Answers already cached
Every DNS answer carries a TTL (time to live), the number of seconds a device may reuse it. If a device resolved a domain before it was classified, it can reuse that answer until the TTL expires. Protective DNS acts on new lookups, so a short cached window can slip through.
Legitimate services abused for C2
Attackers host C2 on popular cloud storage and collaboration platforms. Blocking the whole domain would break legitimate work for everyone. These cases need endpoint detection or traffic inspection, not a domain-level block. Malware can also hide data inside DNS queries themselves. Our guide to DNS tunnelling detection covers how to spot that pattern.
False positives
Any detection that acts before a blocklist will occasionally flag a legitimate new site. Keep an allow list per network, review blocked queries weekly, and give staff a clear route to report a wrong block. Guest networks usually tolerate stricter rules than you expect, because guests rarely visit domains that are days old.
Does DNS filtering replace antivirus?
The two controls work at different points. DNS blocking cuts the callback and the download host before contact. Endpoint protection inspects whatever still arrives, through USB, email attachments or IP-only traffic. Together they shrink both the number of threats reaching the device and the time an infection stays useful to an attacker.
How to verify that blocking works
Verification is a short routine, and you should repeat it after every network change.
- Test resolution. Query a test domain your protective DNS provider publishes, using
digornslookup, from a device on each network and VLAN. - Confirm the resolver. Check that DHCP on every segment hands out the protective resolver, not a fallback public one.
- Try a bypass. From a test device, attempt DNS over HTTPS to a public resolver and confirm your firewall stops it.
- Read the logs. Confirm the blocked query appears with the right network, time and category.
Some resolvers return Extended DNS Errors, defined in RFC 8914, with codes such as Blocked and Filtered. These make blocks visible in diagnostic tools. For Shield's configuration steps, see how Shield works.
How does Shield block malware domains on your network?
Shield is Purple's protective DNS service. It runs standalone on any WiFi or bolts onto Purple Access, with a policy per network and VLAN. You can hold a payment segment to a strict allow list while the guest network blocks malware, phishing and newly registered threats.
Shield is hardware-agnostic. It needs no new appliance, and it works across Cisco Meraki, HPE Aruba, Ruckus, Juniper Mist, Ubiquiti UniFi, Cambium, Extreme and Fortinet estates. You change where DNS points, not the hardware in your ceiling.
Resilience decides whether a DNS control is safe to deploy. Shield fails closed, so a lookup it cannot check is never waved through unfiltered. It runs across multiple data centers to keep that failure mode rare. The wider Purple platform carries 500 million logins a year at a 99.999% uptime SLA.
- Up to 10 daysearlier detection than standard blocklistsPurple data
- ~241,000threats blocked a weekSt George's Healthcare NHS Trust
- 99.999%platform uptime SLA500 million logins a year
- 80,000+venues in 90 countriesPurple data
Shield sits in the same category as Cisco Umbrella. Both apply security policy at the resolver, before any connection opens. Shield's difference is placement. When you bolt it onto Purple Access, it shares a platform with your guest and staff WiFi across 80,000+ venues in 90 countries. For the wider security picture, read our enterprise WiFi security guide and the protective DNS overview.
The concept bites hardest on networks full of devices you do not control. Guest WiFi, device segments and shared healthcare networks are the places where a C2 callback otherwise succeeds unseen. Phishing domains follow the same lifecycle as malware hosts, so read stopping phishing at the DNS layer next.
Key terms
- Command and control (C2)
- The channel a compromised device uses to receive instructions and return stolen data. MITRE ATT&CK catalogs it as a distinct tactic, TA0011. Most malware locates its C2 server by domain name rather than a fixed IP address. C2 callbacks are the main target of DNS-layer blocking. A device repeatedly asking for a blocked C2 domain is probably infected, and the resolver log identifies the network and time.
- Protective DNS
- A recursive resolver that checks every lookup against threat intelligence and policy, then refuses to answer for malicious domains. The UK NCSC runs this model for the public sector through its PDNS service, and CISA offers a protective DNS resolver to US federal civilian agencies. It acts at resolution, before any TCP or TLS session exists, which makes it the one control you can apply to every device on guest WiFi.
- Domain Name System (DNS)
- The naming system defined in RFC 1034 and RFC 1035 that turns a domain name into an address. Devices send queries to a recursive resolver, which performs lookups through to the domain's authoritative servers. Every device must resolve names before it can use them, so the resolver your network hands out is the natural place to enforce security policy.
- Recursive resolver
- The DNS server that receives a device's query and resolves it on the device's behalf, following the resolution process described in RFC 1034 and RFC 1035. Networks usually assign it through DHCP. Verifying that DHCP on every segment hands out the protective resolver, not a fallback public one, is a core step in confirming blocking works.
- Domain generation algorithm (DGA)
- A technique that produces long lists of candidate domain names, which malware tries until one resolves. MITRE ATT&CK catalogs it as technique T1568.002 under Dynamic Resolution. DGA domains appear faster than blocklists can list them. Behavioral detection catches them by scoring lexical structure, because DGA output tends to look random.
- Fast flux
- An evasion technique that rotates the IP addresses behind a single domain name to frustrate address-based blocking. Fast flux weakens firewall rules built on known IPs, which is one reason to judge the domain itself at resolution time.
- Response Policy Zones (RPZ)
- A resolver feature, described in IETF draft-vixie-dnsop-dns-rpz, that rewrites DNS answers according to policy zones loaded from threat feeds. It is the usual basis of a DNS firewall. An RPZ-based DNS firewall is only as current as its last feed update, so it misses domains in the gap before listing unless detection is added on top.
- Newly registered domain blocking
- A policy rule that blocks any domain registered within a set number of days, regardless of whether it appears on a blocklist. It catches campaigns that behavioral scores miss, at the cost of occasional friction with legitimate new sites. It suits staff and device networks better than open guest networks.
- DNS over HTTPS (DoH)
- A protocol defined in RFC 8484 that carries DNS queries inside encrypted HTTPS sessions to a chosen resolver. Browsers and apps sending DoH to their own resolver skip your protective DNS policy. Block known public DoH endpoints at the firewall to close this bypass.
- DNS over TLS (DoT)
- A protocol defined in RFC 7858 that encrypts DNS queries over a TLS connection between client and resolver. Like DoH, it protects privacy on hostile networks but can route lookups around your policy if devices point it at an outside resolver.
- Time to live (TTL)
- The value carried in every DNS answer that sets the number of seconds a device may reuse that answer before querying again. A device that resolved a domain before it was classified can reuse the cached answer until the TTL expires, leaving a short window protective DNS cannot close.
- Extended DNS Errors (EDE)
- A mechanism defined in RFC 8914 that lets a resolver attach error codes, such as Blocked and Filtered, to a DNS response. Resolvers that return EDE make blocks visible in diagnostic tools such as dig, which simplifies verification on each network and VLAN.
- Endpoint detection and response (EDR)
- Software running on a managed device that inspects files, processes, and behavior to detect and respond to malicious activity. EDR covers what DNS blocking cannot, including hard-coded IP traffic and infections already on the device. It cannot be installed on unmanaged guest devices.
Worked examples
You run a venue network where card terminals, cameras and digital signage share infrastructure with staff and guest WiFi. None of those devices can run endpoint software, and the card terminals sit in scope for PCI DSS. How do you protect them?
Place each device type on its own VLAN and point every VLAN at the protective resolver through DHCP. Give the payment segment a strict allow list, so terminals can resolve only the domains they need. The guest network keeps a broader policy that blocks malware, phishing and newly registered threats. This works because DNS is the one control every device must use, whether or not it accepts an agent. Narrowing the domains card terminals can reach reduces the exposure of that PCI DSS segment. Verify the setup by querying a test domain from each VLAN and confirming the blocked query appears in the logs.
You already run a DNS filtering service across several sites and want to move to Shield without disrupting staff or guests. How do you migrate safely?
Migrate network by network. Point one site or VLAN at Shield, recreate its allow and block lists, then compare blocked queries against your current service for two weeks. Once the results match your expectations, move the remaining networks in batches. Because Shield sets policy per network and VLAN, both services can run on different segments during cutover. No hardware changes are needed, since Shield is hardware-agnostic and runs as a cloud overlay. If anything goes wrong, rollback means pointing DNS back to the previous resolver. Repeat the verification routine after each batch.
Frequently asked questions
Does DNS filtering replace antivirus?
No. DNS filtering stops a device reaching malicious domains, while antivirus and EDR inspect files and behavior on the device itself. Protective DNS cuts the C2 callback and the download host, which reduces what endpoint tools must catch. It cannot remove malware already installed or see traffic to hard-coded IP addresses. Run both on managed devices. On guest WiFi, where you cannot install anything, protective DNS is the control you can apply.
Can protective DNS block zero-day domains?
Yes, in part. A zero-day domain has no blocklist entry, so list-only filtering lets it through. Behavioral detection judges the domain on age, structure, hosting and query patterns instead. Purple data shows Shield flags threats up to 10 days before standard blocklists. Detection is probabilistic, so pair it with newly registered domain rules on sensitive segments and endpoint protection on managed devices.
Will Shield work with the WiFi hardware we already own?
Yes. Shield works at the DNS layer, so it does not depend on your access points or controllers. Purple is hardware-agnostic and runs as a cloud overlay on Cisco Meraki, HPE Aruba, Ruckus, Juniper Mist, Ubiquiti UniFi, Cambium, Extreme and Fortinet. You point each network or VLAN at Shield and set its policy. No rip and replace is required.
Do we need new hardware or device agents to deploy it?
No. Shield is a cloud service, so there is no appliance to rack and no software to install on guest devices. Deployment means pointing each network's DNS at Shield, usually through DHCP, and choosing a policy per network or VLAN. That makes it practical for guest WiFi, where devices belong to guests and passengers. The Shield product page covers the setup steps.
Is Shield an extra cost on top of Purple WiFi?
Yes. Shield adds onto Purple Access, and you can also run it standalone on any network, wired or wireless, that does not use Purple. Because it needs no new hardware, you pay for the service, not for appliances or installation visits. Shield is quoted on your sites, networks, users and plan, with a 30-day free trial.
How does Shield compare with Cisco Umbrella?
They sit in the same category: both apply security policy at the resolver, before a connection opens. Shield's difference is placement inside Purple. It runs standalone on any WiFi or bolts onto Purple Access, with a policy per network and VLAN. Shield fails closed and runs across multiple data centers. When you shortlist, compare detection lead time, failure behavior and per-VLAN policy control.
How do we migrate from an existing DNS filtering service?
Network by network. Point one site or VLAN at Shield, recreate its allow and block lists, and compare blocked queries against your current service for two weeks. Then move the remaining networks in batches. Because policy is set per network and VLAN, the two services can run on different segments during cutover. No hardware changes, so rollback means pointing DNS back.
Does protective DNS help with compliance?
Yes, as a supporting control. CISA provides protective DNS guidance to the US public sector, and blocking malicious domains supports malware protection frameworks like SOC 2 Type II expectations. It does not replace device-level anti-malware requirements. DNS logs can identify individuals, so treat them as personal data under CCPA/CPRA. Purple is ISO 27001 certified and CCPA/CPRA compliant.
Sources
- MITRE ATT&CK TA0011: Command and Control
- MITRE ATT&CK T1568.002: Domain Generation Algorithms
- NCSC Protective DNS (PDNS)
- CISA Protective Domain Name System Resolver
- RFC 1034: Domain names, concepts and facilities
- RFC 8484: DNS queries over HTTPS (DoH)
- RFC 8914: Extended DNS Errors
- IETF draft-vixie-dnsop-dns-rpz: DNS Response Policy Zones
