Skip to content
Shield

Decision guideDNS filtering limits

Is DNS filtering enough? Where it ends and what to add

Decide, network by network, whether protective DNS is enough or whether you need a secure web gateway or endpoint layer beside it. You will learn how to close the three gaps DNS filtering leaves at the network: encrypted DNS bypass, lateral movement and direct-to-IP traffic. You will also see which venue networks need...

14 min read

Is DNS filtering enough? Where it ends and what to add

Watch the explainer

Use protective DNS alone when you do not manage the devices and guests sit on their own VLAN. Use DNS plus a secure web gateway when you manage the devices and must inspect full URLs. DNS filtering acts at name resolution, so it never sees paths, payloads or lookups that bypass it via DNS over HTTPS (RFC 8484).

Is DNS filtering enough on its own? The rule of thumb

Start with placement, because placement decides what any filter can see. DNS filtering acts at one moment: when a device asks a resolver to turn a domain name into an IP address. If the domain is malicious or sits in a blocked category, the resolver refuses to answer. No connection opens, so the threat never reaches the device.

That placement is both the strength and the limit. It works on every device that uses the network's resolver, including cell phones you will never manage. It needs no agent and no new hardware, only a DNS change on the network.

The resolver sees the domain and nothing else. It never sees the path after the slash or the file inside a download. It also misses traffic that never asks it a question.

So the rule of thumb has two halves:

  • DNS filtering is enough as the threat and content layer on networks of devices you do not manage. That holds once you close three gaps at the network: encrypted DNS bypass, lateral movement and direct-to-IP traffic.
  • DNS filtering is not enough on its own where you manage the devices and a named rule requires full URLs, files or data loss controls.

The diagram shows where each layer sits on a single request. Our guide to DNS filtering limits covers the wider picture.

DNS querythreat or category matchcleanmanaged deviceunmanaged devicepasses inspectionbypassunfilteredDevice looks up adomainGuest phone, stafflaptop or IoT sensorShield resolvesthe queryPolicy set per networkand VLANDoH, DoT or VPNlookupSkips the networkresolver unless blockedBlocked beforeconnectingMalware, phishing or ablocked categoryAllowed: IPaddress returnedDevice opens a TLSconnectionGateway orendpoint inspectsFull URL and files,managed devices onlySite loadsContent reaches thedevice
Where DNS filtering sits on a single web request, and what it cannot see

A device's DNS query goes to Shield, which blocks or allows it, while encrypted DNS and VPN lookups bypass the resolver unless the firewall stops them.

Connections

  1. Device looks up a domain to Shield resolves the query (DNS query)
  2. Shield resolves the query to Blocked before connecting (threat or category match)
  3. Shield resolves the query to Allowed: IP address returned (clean)
  4. Allowed: IP address returned to Gateway or endpoint inspects (managed device)
  5. Allowed: IP address returned to Site loads (unmanaged device)
  6. Gateway or endpoint inspects to Site loads (passes inspection)
  7. Device looks up a domain to DoH, DoT or VPN lookup (bypass)
  8. DoH, DoT or VPN lookup to Site loads (unfiltered)

Which layer does your network need? The decision in one chart

Two questions settle most cases: who owns the device, and what shares the network with it. Answer the chart for each network or VLAN, not for the whole property. A hotel's guest network and its back office network usually land in different places.

Is DNS filtering enough for this network?

See every path as a list
Do you manage the devices on this network?
  • No, unmanaged: Do payment, staff or building systems share the same physical network?
    • Yes: Shield with a policy per VLAN Move payment and staff systems onto their own VLANs, then give each VLAN its own Shield policy. DNS filtering does not stop lateral movement; segmentation does.
    • No: Shield plus client isolation Protective DNS on the guest network, port 853 blocked at the firewall and client isolation on. DNS is the layer you can enforce on devices you do not own. Fail open vs fail closed
  • Yes, managed: Does a policy or regulation require you to inspect full URLs or downloaded files?
    • Yes: Shield plus a secure web gateway Keep Shield as the first block and add TLS inspection on managed devices for the paths and files DNS cannot see. When you need the heavier tool
    • No: Do these devices regularly work away from your networks?
      • Yes: Shield plus endpoint protection Shield covers every network you point at it. Endpoint controls cover the device on a home or hotel router you do not run.
      • No: Shield on its own Protective DNS with threat and content categories, firewall rules against encrypted DNS bypass, and the endpoint protection you already run.

Every outcome keeps protective DNS in place. The question is never whether to run DNS filtering, but what to add beside it. DNS is the only layer that reaches unmanaged devices, so it stays first everywhere.

Which setup fits your venue? Scenarios and recommendations

The table maps common venue networks to a recommendation and the condition behind it. Where a venue runs several networks, read one row per network.

DNS filtering gaps by venue network: what to run and why Data as of 2026-06-10
ScenarioRecommendationWhy
Hotel guest WiFiShield on the guest VLAN, client isolation, port 853 blockedGuest devices take no agent, so DNS is the layer you can enforce
Bar or restaurant with card terminalsShield per VLAN, terminals on their own segmentPCI DSS scope follows segmentation; DNS cannot stop lateral movement
Airport passenger WiFiShield on the passenger network, stricter policy on operations VLANsHigh volumes of unmanaged devices beside operational systems
Stadium or event attendee WiFiShield with threat categories, client isolationHigh volumes of unmanaged devices require fast threat filtering at the DNS layer
Hospital patient and visitor WiFiShield on visitor networks, separate policy for clinical VLANsSt George's Healthcare NHS Trust blocks around 241,000 threats a week with Shield
Retail shopper WiFiShield with family-safe content categoriesShoppers bring their own devices; content on your network carries your name
Retail staff tablets, managedShield plus MDM and endpoint protectionYou own the device, so device controls add payload inspection
K-12 schools and collegesShield plus full-URL filtering and monitoring on managed devicesCIPA filtering and monitoring standards ask for monitoring as well as blocking
Multi-Family or Student Housing residentsShield threat categories per resident networkResidents expect open internet; threat blocking protects without policing content
Corporate office with regulated dataShield plus a secure web gateway with TLS inspectionData loss rules need file and path visibility on managed laptops
Coworking membersShield threat categories, members separated by VLANMember devices are unmanaged and must not reach each other
Recommendations assume Shield runs standalone or on Purple Access, with a policy per network and VLAN.

Two patterns stand out. Every network of visitor-owned devices lands on Shield plus network controls, because you cannot install anything on a guest's cell phone. The heavier layers only appear where you own the device and a rule demands deeper visibility.

What does protective DNS with network controls cover?

This is option one, and the right answer for most guest, resident, and visitor networks. Purple Shield is protective DNS and content filtering for every network. It runs standalone or on Purple Access, with a separate policy per network and VLAN.

What Shield covers on its own

Shield blocks malware, phishing, and other threat domains before a connection opens. It applies content categories, so a family hotel can keep adult content off guest WiFi while staff stay unrestricted. Shield identifies threat domains early to stop malicious connections, reducing unnecessary network traffic and lookup overhead.

Shield also fails closed, so an outage never leaves a network unfiltered. A filter that fails open quietly switches protection off at the worst moment. Our guide to fail open vs fail closed explains the trade-off.

The three gaps you close at the network

DNS filtering leaves three gaps that your existing firewall and controller can close:

  1. Encrypted DNS bypass. DNS over HTTPS hides lookups inside ordinary HTTPS on port 443. DNS over TLS (RFC 7858) uses port 853. Block port 853 outbound, redirect stray port 53 traffic to your resolver, and block known DoH endpoints. Our guide to DNS over TLS and port 853 walks through it.
  2. Lateral movement. A DNS filter does not stop one infected device probing another on the same subnet. Client isolation stops guests reaching each other. VLAN segmentation keeps guests away from registers, cameras, and building systems. On multi-family sites, Purple's Identity-Based Networks give each resident a private network by design.
  3. Direct-to-IP traffic. Malware that connects to a hard-coded IP address never asks the resolver anything. Egress rules on your firewall close the obvious routes.

A guest on a personal VPN resolves names inside the tunnel, so no network-side filter sees those lookups. Treat that as a matter for your acceptable use terms, not a filter setting.

Worked scenario: a 200-room hotel

Situation. A 200-room hotel runs guest WiFi, a staff network, and card terminals on the same switches. A security review flagged two risks. Guest WiFi had no threat filtering, and payment devices were reachable from the guest subnet.

What was done. The hotel moved its card terminals onto their own VLAN and turned on client isolation for guests. It pointed the guest and staff VLANs at Shield, each with its own policy. Guests got threat and adult content blocking; staff got threat blocking only. Port 853 was blocked at the site firewall.

Outcome. One resolver change per VLAN, no new hardware, and no agent on any guest device. The card terminals left the guest network's reach, which narrows the systems in PCI DSS scope. Every guest lookup now passes a threat check before a connection opens.

Worked scenario: a healthcare network

Situation. St George's Healthcare NHS Trust needed threat protection across a busy hospital estate. Patients and visitors bring devices the trust will never manage.

What was done. The trust runs Shield, which blocks threat domains at the DNS layer before any connection opens.

Outcome. Around 241,000 threats are blocked every week. Each one is a connection that never opened on a hospital network.

When do you need a secure web gateway on top of DNS filtering?

Option two adds a layer that sees inside the connection. A secure web gateway sits in the traffic path as a proxy. With TLS inspection, it decrypts HTTPS, reads the full URL and file, then re-encrypts the traffic.

What a gateway sees that DNS cannot

DNS sees drive.google.com; a gateway sees the specific shared folder. That difference matters when one domain hosts both the tool you need and the content you must stop. It also matters for data loss rules, file-type blocking and scanning inside downloads. Our guide on why DNS filtering cannot see full URLs covers the detail.

Why it only works on devices you manage

TLS inspection needs a root certificate the device trusts, and usually an agent or proxy setting to route traffic. You can push both to managed laptops through MDM. You cannot push them to a guest's cell phone. Some apps pin their certificates and break under inspection, so expect an exception list to maintain.

That makes the gateway a managed-device layer, never a guest-network layer. Where you need one, keep Shield in front of it. Shield stops known threat domains before the gateway sees them, so the gateway has less to inspect.

How Cisco Umbrella and Cloudflare Gateway fit the picture

Cisco Umbrella and Cloudflare Gateway both start at the DNS layer and extend into proxy inspection. Cisco Umbrella adds a secure web gateway in its higher tiers. Cloudflare Gateway applies HTTP policies to traffic routed through its WARP client or a proxy endpoint. TLS decryption there needs a certificate on the device. In both cases the proxy features depend on reaching the device.

So the comparison turns on the same condition as this guide. On networks of unmanaged devices, the DNS tier is the part any provider can enforce. Shield is built for those networks: a policy per network and VLAN, standalone or on Purple Access, and fail-closed behavior. Our guide to DNS filtering vs secure web gateway sets out when the heavier tool earns its place.

What does each option cost to run?

The license is rarely the biggest cost. Effort is: who configures each layer, what breaks and what needs maintaining every month.

Layer Gap it closes Setup effort Ongoing effort Works on unmanaged devices
Protective DNS (Shield) Threat domains and blocked categories A DNS change per network or VLAN Review block logs and category policy Yes
Firewall rules DoT on port 853, rogue resolvers on port 53 A few rules per site firewall Refresh the DoH endpoint list Yes
Client isolation and VLANs Lateral movement between devices VLAN design and controller settings Re-check each time a system joins Yes
Secure web gateway with TLS inspection Full URLs, files and data loss Agent or proxy plus a root certificate per device Certificate renewal, app exceptions, agent updates No
Endpoint protection or MDM Payloads and off-site devices Enroll every device Patching and per-device licenses No

Shield is priced by tailored quote, and you can run it free for 30 days before you commit. Because it is a DNS change, you can trial it on one network without touching the rest.

A gateway's cost sits mostly in operations. Certificates need distributing, agents need updating and the exception list grows each time an app breaks. Segmentation costs design time once, then little after. If you add one control to DNS filtering this quarter, make it client isolation on guest networks.

What changes the answer?

The rule of thumb holds until one of these conditions changes. Each one pushes the recommendation in a specific direction.

  • You start managing the devices. A staff network of enrolled laptops can take an agent and a certificate. That opens the gateway and endpoint options.
  • Payment or building systems join the network. Segmentation moves from good practice to essential. PCI DSS scope follows the cardholder data environment, so isolate it before you tune any filter.
  • A rule asks for monitoring, not only blocking. K-12 schools and colleges work to federal and state compliance standards, which ask for both. Data loss rules in finance teams ask for file visibility. Either points to a gateway on managed devices.
  • Devices leave your sites. Shield applies a policy to each network you point at it. A laptop on a home router sits on none of your networks, so add endpoint controls for off-site work.
  • Encrypted DNS appears in your logs. If resolver queries fall while bandwidth holds steady, check port 853 and DoH endpoints first.
  • Downtime tolerance is zero. Ask any provider what happens when the filter fails. Shield fails closed; a fail-open filter leaves the network open for as long as the outage lasts.

Density bends the answer at events. A stadium network carries thousands of attendee cell phones at once, all unmanaged. Shield reduces DNS query overhead and blocks unwanted traffic before connections open, which frees capacity where it is scarcest.

What should you deploy by default?

Which approach fits your network?

Shield with network controls

Fits when

  • You do not manage the devices
  • Guests, residents or fans bring their own cell phones
  • You need protection this quarter with no new hardware

Not when

  • A rule requires full-URL or file inspection on managed devices

Shield plus a secure web gateway

Fits when

  • You manage every device on the network
  • You can deploy an agent and a root certificate
  • A named rule requires path or file visibility

Not when

  • The network carries guest or resident devices

Shield plus endpoint protection

Fits when

  • Managed laptops work off site
  • You already run MDM

Not when

  • The devices belong to visitors

Our recommendation: Start every network on Purple Shield with port 853 blocked and guests isolated. Add a gateway or endpoint layer only on managed devices, where a named rule requires it.

The default for almost every venue network is protective DNS with network controls. It reaches every device, needs no new hardware and fails closed. Add heavier layers only where you own the device and can name the rule that demands them.

Your first action: pick one guest network and start the 30-day free trial. Block port 853 on that site's firewall. Then compare a week of Shield's block log with what your current setup reports. See how Shield works for how it deploys. Our enterprise WiFi security guide covers the wider controls around it.

Key terms

Protective DNS
DNS filtering applied at name resolution under the Domain Name System defined in RFC 1034 and RFC 1035: the resolver refuses to return an IP address for domains that match threat intelligence or blocked content categories, so no connection opens. Purple Shield is protective DNS with a policy per network and VLAN. It needs only a resolver change, which is why it is the layer you can enforce on guest, resident, passenger and fan devices.
DNS over HTTPS (DoH)
Specified in RFC 8484, DoH carries DNS queries and responses inside HTTPS messages, normally on TCP port 443, so lookups look like ordinary web traffic and can bypass the network resolver. A browser using DoH resolves names without touching your resolver. Blocking known DoH endpoints at the firewall closes this bypass on any device.
DNS over TLS (DoT)
Specified in RFC 7858, DoT encrypts DNS between client and resolver using TLS over a dedicated port, 853, preventing on-path observation or modification of queries. Because DoT uses a fixed port, blocking outbound port 853 at the site firewall is one of the cheapest controls to pair with DNS filtering.
Secure web gateway
A proxy placed in the traffic path that applies policy to full HTTP and HTTPS requests, including URL paths, file types and download content, rather than to the domain name alone. A gateway sees the specific shared folder where DNS only sees the domain. It needs an agent or proxy setting on each device, so it belongs on managed devices only.
TLS inspection
Interception of Transport Layer Security sessions (TLS 1.3 is specified in RFC 8446) in which a proxy decrypts traffic, inspects it and re-encrypts it, which requires the client to trust the proxy's root certificate. You can push a root certificate to managed laptops through MDM but never to a guest's phone. Apps that pin certificates break under inspection, so plan an exception list.
VLAN
A virtual local area network defined by IEEE 802.1Q, which tags Ethernet frames to separate one physical network into isolated logical broadcast domains. Moving payment, staff and building systems onto their own VLANs stops guest traffic reaching them and lets each VLAN carry its own Shield policy.
Client isolation
A wireless controller or access point setting that blocks layer 2 traffic between clients associated to the same SSID, so devices on one guest network cannot communicate directly. DNS filtering cannot stop one infected device probing another on the same subnet. Client isolation is the first control the guide recommends adding to guest networks.
PCI DSS
The Payment Card Industry Data Security Standard, published by the PCI Security Standards Council, sets security requirements for any environment that stores, processes or transmits cardholder data; network segmentation can reduce the systems in scope. Bars, restaurants and hotels running card terminals beside guest WiFi narrow PCI DSS scope through segmentation. DNS filtering supports compliance but does not satisfy it.
Cardholder data environment
The PCI DSS term for the people, processes and systems that store, process or transmit cardholder data, plus any system connected to them, which together define assessment scope. PCI DSS scope follows the cardholder data environment, so isolate terminals and registers on their own segment before tuning any filter.
Fail closed
A failure mode in which a security control blocks or keeps enforcing policy when it loses service, as opposed to fail open, where traffic continues without enforcement until service returns. Shield fails closed, so an outage never leaves a network unfiltered. Ask any provider which way its filter fails before you sign.
Lateral movement
Movement by an attacker or malware from one compromised host to other hosts on the same network, typically by scanning and probing peers on the local subnet. One of the three gaps DNS filtering leaves. You close it with client isolation and VLAN segmentation, or Identity-Based Networks on multi-family sites.
Mobile device management (MDM)
Platforms that enroll and centrally configure organization-owned devices, including pushing certificates, proxy settings, agents and security policy. MDM is what makes TLS inspection and endpoint protection possible. Without it, as on every visitor-owned device, DNS carries the load.

Worked examples

A 200-room hotel runs guest WiFi, a staff network and card terminals on the same switches. A security review found no threat filtering on guest WiFi and payment devices reachable from the guest subnet. What should the hotel change?

The hotel moved its card terminals onto their own VLAN and turned on client isolation for guests. It pointed the guest and staff VLANs at Shield, each with its own policy: guests got threat and adult content blocking, staff got threat blocking only. Port 853 was blocked at the site firewall to stop DNS over TLS bypass. The change took one resolver setting per VLAN, no new hardware and no agent on any guest device. Segmentation took the terminals out of the guest network's reach, which narrows the systems in PCI DSS scope, and every guest lookup now passes a threat check before a connection opens.

St George's Healthcare NHS Trust needed threat protection across a busy hospital estate where patients and visitors bring devices the trust will never manage. Is DNS filtering enough there?

The trust runs Shield, which blocks threat domains at the DNS layer before any connection opens. That choice fits the decision chart: the devices are unmanaged, so no agent or root certificate can be installed and a secure web gateway cannot reach them. Protective DNS is the layer that applies to every device using the network's resolver. The guide recommends a separate Shield policy for clinical VLANs, kept apart from visitor networks. The outcome is around 241,000 threats blocked every week, each one a connection that never opened on a hospital network.

A school or college in England runs managed devices and must meet the Department for Education filtering and monitoring standards. Can it rely on DNS filtering alone?

No. The DfE standards ask for monitoring as well as blocking, which is a named rule demanding deeper visibility than DNS provides. DNS sees only the domain, never the path after the slash or the file in a download. Because the school owns the devices, it can deploy an agent and a root certificate through MDM. The recommendation is Shield as the first block, plus full-URL filtering and monitoring on managed devices. Shield stops known threat domains before the heavier layer sees them, so the gateway has less to inspect.

Frequently asked questions

Is DNS filtering enough for PCI DSS compliance?

No: DNS filtering supports PCI DSS but does not satisfy it. PCI DSS scope follows the cardholder data environment, and segmentation is the main lever on scope. Put card terminals and POS registers on their own VLAN and isolate guests. Then give each VLAN its own Shield policy. The filter cuts malware and phishing risk; segmentation keeps guest traffic away from payments.

Does Shield work with the network hardware we already run?

Yes. Shield works at the DNS layer, so it needs a DNS setting, not new hardware. That covers the platforms Purple works with, including Cisco Meraki, HPE Aruba, Ruckus, Juniper Mist, Ubiquiti UniFi, Cambium, Extreme and Fortinet. You point each network or VLAN at Shield and assign a policy. It runs standalone or on Purple Access, the same way on every site.

How much does Shield cost?

Shield is priced by tailored quote for your estate, with a 30-day free trial first. You can trial it on a single network and compare its block log with your current filter. There is no agent to deploy on devices and no hardware to buy. Ask for a quote once the trial shows what Shield catches on your own traffic.

How does Shield compare with Cisco Umbrella and Cloudflare Gateway?

Shield does the same DNS-layer job and is built for networks full of devices you do not manage. Cisco Umbrella and Cloudflare Gateway extend DNS filtering into proxy inspection, which needs an agent or certificate on each device. On guest, resident and passenger networks you cannot deploy either, so the DNS tier carries the load. Shield adds a policy per network and VLAN and fails closed.

Can guests get around DNS filtering?

Some can, through DNS over HTTPS, DNS over TLS or a personal VPN. Each resolves names away from your resolver. Block outbound port 853 and known DoH endpoints at the firewall, and redirect stray port 53 traffic to Shield. That closes the common routes on any device. A VPN tunnel stays opaque to every network filter, so cover it in your acceptable use terms.

How hard is it to migrate from another DNS filter?

Migration is a resolver change and a policy rebuild, not a hardware project. Recreate your categories and allow lists in Shield, point one network at it and compare results during the 30-day free trial. Once the block logs match or improve, move the remaining networks and VLANs one at a time. Keep the old filter live on untouched networks until each cutover is confirmed.

What happens to filtering if Shield goes down?

The network stays filtered. Shield fails closed, so an outage never leaves a network unfiltered, which matters on guest WiFi where nobody watches each device. A fail-open filter keeps traffic flowing during an outage but drops every block until service returns. Ask any provider which way its filter fails before you sign. Our guide to [fail open vs fail closed](guide:dns-filtering-fail-open-vs-fail-closed) sets out the trade-off.

Sources

Decide what your network reaches, free for 30 days

Tell us your networks and who uses them, and we map a policy to each. Point one at Shield and see what it stops for 30 days.

Tell us your networks. We map a policy to each one.