WalkthroughProtective DNS
Deploying protective DNS per SSID: DHCP, resolvers and policy mapping
You will be able to deploy protective DNS per SSID on the access points you already own, by pointing each VLAN's DHCP option 6 at Purple Shield and mapping each network to one policy. You will also close DNS bypass routes, verify every SSID with three tests and fix the failures that most often break the SSID-to-policy...
Watch the explainer
Deploy protective DNS per SSID in eight steps on the access points you already own. Each SSID sits on its own VLAN, so you point that VLAN's DHCP option 6 (RFC 2132) at Purple Shield and register the network against a policy. Shield then applies that policy to every lookup, before a connection to a malicious site opens.
What will you have when protective DNS runs per SSID?
Filtering happens at the DNS lookup, in the cloud, before a device opens any connection. Nothing sits in the traffic path and nothing installs on guest devices. Because of that placement, the change lives in DHCP rather than in new hardware.
When you finish, four things are true:
- Every SSID hands out Shield resolvers and nothing else.
- Each SSID maps to exactly one Shield policy, such as guest, staff, payments or IoT.
- Devices that hard-code a public resolver hit a firewall block instead of bypassing the filter.
- You hold a test result for each SSID that shows a blocked category blocked and an allowed one resolved.
Per-SSID policy matters because audiences differ. A hotel guest network needs malware and adult content blocked. Back of house needs threat protection without breaking supplier portals. A payments VLAN should reach only what the terminals need. A single policy across all three either over-blocks staff or under-protects guests.
What do you need before you start?
Gather these before you open the Shield console. The SSID-to-VLAN map decides whether everything after it works, so build it first.
| Prerequisite | Why it matters | Where you get it |
|---|---|---|
| Shield account | Holds your policies and the network-to-policy mapping | Standalone or on Purple Access; 30-day free trial |
| SSID-to-VLAN map | DHCP scopes serve subnets, so policy follows the VLAN, not the SSID name | Your WiFi controller or cloud dashboard |
| Admin access to every DHCP server | Option 6 is set wherever DHCP runs, which may not be the access point | Access point, controller, gateway, firewall or Windows Server |
| Firewall rule access | Stops devices bypassing Shield with hard-coded DNS | The firewall or gateway at each site |
| Captive portal domain list | The portal must still resolve before a guest logs in | Your captive portal settings |
| Signed-off policy design | Decides what each audience can reach | IT security with venue operations |
| Shortened DHCP lease | Clients pick up new DNS servers only when they renew | Each scope, changed a day before the window |
| Test device per SSID | Proves the result on every network | Any laptop you can join to each SSID |
Shorten the lease the day before you change anything. Under RFC 2131, clients renew halfway through the lease by default. A 24-hour lease therefore leaves some devices on old DNS servers for up to 12 hours.
How does an SSID map to a Shield policy?
The SSID never talks to Shield directly. The chain runs SSID to VLAN, VLAN to subnet, subnet to DHCP scope, scope to resolver, resolver to policy. Break any link and the policy is wrong or missing.
DHCP (Dynamic Host Configuration Protocol) gives each device its address and its DNS servers in one lease. Option 6 carries the DNS server list. A VLAN (virtual LAN) keeps each SSID's traffic on its own subnet. Each DHCP scope serves one subnet, so the scope is where per-SSID DNS is decided.
Shield then identifies which registered network a query came from and applies the policy you assigned to that network. The how Shield works page covers the detail.
A device joins an SSID, receives Shield resolvers from that VLAN's DHCP scope, and Shield applies the network's policy to allow or block each lookup.
Connections
- Device joins the guest SSID to DHCP scope for that VLAN (DHCP request)
- DHCP scope for that VLAN to DNS query sent to Shield (Lease with DNS servers)
- DNS query sent to Shield to Shield matches the network
- Shield matches the network to Allowed: address returned (Allowed category)
- Shield matches the network to Blocked: block response (Threat or blocked category)
Cisco Umbrella and Cloudflare Gateway follow the same pattern at network level. You point DHCP at the provider's resolvers, then register the network so the right policy applies. Shield's difference is where it sits. It runs standalone, or on Purple Access, where each network's and VLAN's policy sits in the same platform as the WiFi it protects.
How do you deploy protective DNS per SSID, step by step?
Work one site at a time and finish each check before the next step. Steps 1 to 3 happen in planning and in the Shield console. Steps 4 to 8 happen on your network.
Deploy protective DNS per SSID
Step 1: Map every SSID to its VLAN and subnet
List each SSID, its VLAN ID, its subnet and the device that runs DHCP for it.
Check it worked: Every SSID has one VLAN and one named DHCP server, with no two SSIDs sharing a VLAN.
Step 2: Create one Shield policy per audience
In the Shield console, build a policy for each audience: guest, staff, payments, IoT.
Check it worked: Each policy lists the categories and threat protections you signed off.
Step 3: Register each network and assign its policy
Add the network behind each SSID in Shield and attach the matching policy.
Check it worked: The network list shows one policy per SSID and no unassigned networks.
Step 4: Set DHCP option 6 on each scope
Replace the DNS servers in each SSID's scope with the Shield resolvers. Remove any public or ISP fallback.
Check it worked: The scope lists Shield resolvers and nothing else.
Step 5: Close the bypass routes
Allow DNS from each subnet to Shield only. Block outbound port 53 and 853 elsewhere, and align IPv6 DNS.
Check it worked: A query sent straight to a public resolver from a test device times out.
Step 6: Renew a lease on each SSID
Forget the network on a test laptop, rejoin and read the DNS servers it received.
Check it worked:
ipconfig /all(Windows) orscutil --dns(macOS) shows only Shield resolvers.Step 7: Verify the policy per SSID
Resolve one domain in a blocked category and one in an allowed category on every SSID.
Check it worked: Blocked domains return the block response, allowed domains resolve, and results differ by SSID as designed.
Step 8: Roll out and review the logs
Repeat steps 4 to 7 site by site, then review Shield query logs for each network.
Check it worked: Every network shows queries under its own name and policy.
Steps 1 to 3: build the map before you touch DHCP
Two SSIDs on one VLAN share one scope, so they share one policy. If guest and staff share a VLAN today, split them before you go further. Name each policy after its audience, not its site, so one guest policy serves 40 sites.
Step 4: find the DHCP server that serves the SSID
Where an SSID bridges into a wired VLAN, the access point does not hand out addresses. The firewall, core switch or a Windows server does. Changing DNS on the access point then changes nothing. On a connected Windows device, ipconfig /all shows the DHCP server that answered, which settles the question.
Step 5: close the bypass routes
Devices with hard-coded DNS, or with DNS over TLS on port 853 (RFC 7858), ignore option 6. Add rules in this order on the firewall serving each subnet:
allow udp,tcp 53 from <ssid-subnet> to <shield-resolver-1>, <shield-resolver-2>
deny udp,tcp 53 from <ssid-subnet> to any
deny tcp 853 from <ssid-subnet> to any
DNS over HTTPS (RFC 8484) runs on port 443, so a port rule cannot stop it. On managed staff devices, set the browser's DNS over HTTPS policy through your MDM.
IPv6 is the bypass most teams miss. Router advertisements can carry DNS servers (RFC 8106). On a dual-stack VLAN, a device may prefer the IPv6 resolver your ISP router advertises. Align the router advertisement and DHCPv6 DNS settings with Shield, or stop advertising a third-party resolver.
Worked example: a 200-room hotel
A 200-room hotel runs four SSIDs on HPE Aruba: guest, staff, back of house and a hidden network for in-room TVs. The core firewall serves DHCP for all four VLANs, and every scope handed out the ISP's resolvers.
The team built four Shield policies and dropped lease times to one hour the afternoon before. In one window, they changed option 6 on four scopes and added the port 53 and 853 rules.
The step 6 check caught one fault: the TV VLAN still listed the ISP resolver as secondary. After the fix, all eight tests passed, two per SSID. Guests lost access to blocked categories, while back-of-house staff kept their supplier portals. The protective DNS for hotels guide covers policy design for both sides of the house.
How do you check each SSID gets the right policy?
Run three tests on every SSID from a device that joined after the change. Forget the network first so the device requests a fresh lease rather than reusing a cached one.
# 1. Confirm the DNS servers (Windows, then macOS)
ipconfig /all
scutil --dns | grep nameserver
# 2. Confirm the policy: one blocked, one allowed domain
nslookup <domain-in-a-blocked-category>
nslookup <domain-in-an-allowed-category>
# 3. Confirm the bypass is closed: this should time out
nslookup example.com 8.8.8.8
Pick test domains you know sit in categories each policy blocks or allows. The blocked lookup should return the block response set for that policy. Then open the Shield console and confirm the test queries appear under the network you expect. If they appear under another network, the mapping in step 3 is wrong.
Per-SSID verification
0 of 9 done
Worked example: a 40-store retail estate
A retailer runs shopper WiFi and staff WiFi in 40 stores. Cisco Meraki access points bridge both SSIDs to VLANs, and a Fortinet firewall in each store serves DHCP.
The team piloted two stores first and changed option 6 and the DNS rules on each store firewall. The other 38 stores followed in batches. Each batch had to pass the three tests and show queries under each store's network before the next began.
The pilot surfaced one pattern: staff laptops with DNS over HTTPS enabled resolved blocked test domains. An MDM policy fixed it before batch one. All 40 stores then passed the checklist, with shopper and staff policies separate in every store.
Where does the DNS setting live on your hardware?
The DNS setting lives wherever DHCP runs for that VLAN, not where you configure the SSID. The same rule holds on Ruckus, Juniper Mist, Cambium and Extreme. Setting names vary by firmware, so check your vendor's documentation for the exact label.
Setting option 6 by DHCP location
Access point or gateway DHCP
Cisco Meraki, HPE Aruba and Ubiquiti UniFi can serve DHCP from the access point, controller or gateway. Where they do, set the DNS servers in that SSID's or network's DHCP settings.
- Confirm the SSID is in a mode where the access point or gateway hands out addresses
- Remove any default upstream or ISP resolver
- Apply the change to every site template that carries the SSID
Firewall DHCP
Fortinet and similar firewalls often run DHCP per VLAN interface. Set the DNS servers on each interface's DHCP server, then add the outbound DNS rules on the same device.
- One interface per SSID VLAN
- Rules sit above any general outbound allow
Windows Server DHCP
Set option 006 at scope level for each scope that serves an SSID. Scope options override server options, and reservation options override both.
Set-DhcpServerv4OptionValue -ScopeId <guest-scope-id> -DnsServer <shield-resolver-1>,<shield-resolver-2>
Get-DhcpServerv4OptionValue -ScopeId <guest-scope-id>ISC DHCP
Set the DNS servers inside each subnet declaration, not globally.
subnet <guest-subnet> netmask <mask> {
option domain-name-servers <shield-resolver-1>, <shield-resolver-2>;
}Microsoft documents the cmdlet in its DHCP Server PowerShell reference.
What goes wrong most often, and how do you fix it?
Most failures trace back to one broken link in the SSID-to-policy chain. Match the symptom to the cause.
| Symptom | Likely cause | Fix |
|---|---|---|
| Filtering works intermittently | Public or ISP resolver left as secondary | List Shield resolvers only in option 6 |
| No change after editing the access point | DHCP runs upstream on the firewall or a server | Find the real DHCP server (step 4) and edit it there |
| Some devices still use old DNS | Leases not yet renewed | Shorten lease time, then forget and rejoin |
| Dual-stack devices bypass Shield | IPv6 router advertisements carry another resolver | Align IPv6 DNS with Shield or stop advertising the other resolver |
| Blocked domains resolve on staff laptops | Browser DNS over HTTPS | Set the DNS over HTTPS policy through your MDM |
| Guest and staff get the same policy | Both SSIDs bridge to one VLAN | Split onto separate VLANs, then map each network |
| Captive portal fails to load | Portal domains blocked by the guest policy | Add the portal domains to the policy's allow list |
| One device ignores the scope change | A DHCP reservation carries its own DNS option | Update or remove the reservation-level option |
When a step fails and the table does not explain it, start from the evidence. The Shield query logs show which network a query arrived under and which policy answered it. The DNS filtering logs and reporting guide explains what to record and how to read it. For a site-by-site audit, work through the DNS security checklist for venues.
Where this guide sits in protective DNS
This guide covers placement and mapping: getting the right policy onto the right SSID and proving it. The Protective DNS pillar covers why DNS-layer filtering belongs on every network. To measure performance before and after the change, read DNS filtering latency. For the wider picture of WiFi security controls, see the enterprise WiFi security guide.
Key terms
- DHCP option 6
- The Domain Name Server option defined in RFC 2132, DHCP Options and BOOTP Vendor Extensions. It carries the list of DNS server addresses a DHCP server hands to a client in its lease; Windows Server labels it option 006. You meet it in each SSID's DHCP scope. Replacing its values with Shield resolvers, and nothing else, is the change that puts protective DNS on an SSID.
- DHCP lease renewal
- RFC 2131, Dynamic Host Configuration Protocol, defines lease timers. By default a client tries to renew with its server at T1, half of the lease duration, and only then picks up changed options such as new DNS servers. A 24-hour lease can leave devices on old DNS servers for up to 12 hours, so you shorten the lease a day before the change window.
- VLAN
- A virtual LAN, specified by IEEE 802.1Q, which tags Ethernet frames to separate traffic into distinct broadcast domains. Each VLAN normally maps to its own IP subnet and DHCP scope. Each SSID needs its own VLAN. If guest and staff bridge into one VLAN, they share one scope and therefore one Shield policy.
- DHCP scope
- The address pool and option set a DHCP server, as defined in RFC 2131, uses for one subnet. On Windows Server, scope options override server options, and reservation options override both. The scope is where per-SSID DNS is decided. Set option 6 at scope level, not globally, and check reservations that carry their own DNS option.
- Protective DNS
- DNS-layer filtering that checks each lookup against threat and category policy at the resolver and returns a block response before a device opens a connection. It needs no inline appliance and no software on guest devices. Purple Shield delivers it from the cloud, so deploying it per SSID is a DHCP and firewall change rather than a hardware project.
- DNS over TLS (DoT)
- RFC 7858 specifies DNS over Transport Layer Security, which encrypts DNS queries between client and resolver over a dedicated port, TCP 853. Devices using DoT ignore option 6, so you block outbound TCP 853 from each SSID subnet to stop them bypassing Shield.
- DNS over HTTPS (DoH)
- RFC 8484 specifies DNS queries carried inside HTTPS on port 443, indistinguishable by port from normal web traffic. A firewall port rule cannot stop DoH, so on managed staff laptops you set the browser DNS over HTTPS policy through your MDM.
- IPv6 router advertisement DNS options
- RFC 8106 defines the Recursive DNS Server (RDNSS) and DNS Search List options carried in IPv6 router advertisements, letting a router tell hosts which resolvers to use without DHCPv6. On a dual-stack VLAN, a device may prefer an IPv6 resolver the ISP router advertises. Align router advertisement and DHCPv6 DNS with Shield to close the gap.
- Captive portal
- The web page a guest sees before getting online, presented by the WiFi network until the guest logs in or accepts terms. Portal domains must still resolve under the guest policy. If the portal fails to load after the change, add its domains to that policy's allow list.
- Fail closed
- A design where loss of the filtering service stops resolution rather than letting lookups through unfiltered. A fail-open design falls back to an unfiltered path. Shield fails closed and runs across multiple data centres. Adding a public secondary resolver in option 6 turns the network fail open and lets lookups skip your policy.
- PCI DSS segmentation
- The Payment Card Industry Data Security Standard expects the cardholder data environment to be segmented from other networks, using controls such as VLANs and firewall rules. A payments VLAN with a tight Shield policy adds a DNS control on top of segmentation, and Shield query logs give auditors evidence the control runs.
Worked examples
A 200-room hotel runs four SSIDs on HPE Aruba: guest, staff, back of house and a hidden network for in-room TVs. The core firewall serves DHCP for all four VLANs, and every scope hands out the ISP's resolvers. How do you move all four onto protective DNS?
The team built four Shield policies, one per audience, and dropped lease times to one hour the afternoon before. In one window they changed option 6 on the four firewall scopes and added the port 53 and 853 rules. The step 6 lease check caught one fault: the TV VLAN still listed the ISP resolver as secondary, which would have made that network fail open. After removing it, all eight tests passed, two per SSID. Guests lost access to blocked categories, while back-of-house staff kept their supplier portals, because each SSID mapped to its own policy.
A retailer runs shopper WiFi and staff WiFi in 40 stores. Cisco Meraki access points bridge both SSIDs to VLANs, and a Fortinet firewall in each store serves DHCP. How do you roll out per-SSID protective DNS without disrupting the estate?
Because the firewall serves DHCP, the team changed option 6 and the outbound DNS rules on each store firewall, not on the access points. They piloted two stores first, then rolled the other 38 out in batches. Each batch had to pass the three verification tests and show queries under each store's network in the Shield console before the next began. The pilot showed staff laptops with DNS over HTTPS enabled still resolving blocked test domains, since port rules cannot stop traffic on 443. An MDM policy fixed it before batch one, and all 40 stores then passed with shopper and staff policies kept separate.
Frequently asked questions
How much does Purple Shield cost?
Shield is priced on a tailored quote for your estate, and you can start with a 30-day free trial. It runs standalone on the WiFi you already have, or on Purple Access alongside your guest and staff WiFi. The trial is the practical way to test this guide on your own network. Map one SSID, run the verification checks and review the query logs before you commit.
Will Shield work with the access points I already own?
Yes. Shield works through DHCP and DNS, so it runs on Cisco Meraki, HPE Aruba, Ruckus, Juniper Mist, Ubiquiti UniFi, Cambium, Extreme and Fortinet without new hardware. You need admin access to whichever device serves DHCP for each SSID's VLAN. You also need a firewall that can restrict outbound DNS. Nothing installs on guest devices.
Can guest and staff SSIDs at the same site get different policies?
Yes, as long as they sit on separate VLANs. Shield applies a policy per network and VLAN, and the DHCP scope serving each VLAN decides which resolvers a device receives. Two SSIDs bridged into one VLAN share a scope, so they share a policy. Split them onto separate VLANs first, then register each network in Shield and assign its own policy.
How does Shield compare with Cisco Umbrella and Cloudflare Gateway?
All three filter at the DNS layer and deploy the same way: point DHCP at the resolvers and register each network. Shield's difference is placement. It runs standalone, or on Purple Access, the platform behind WiFi at 80,000+ venues in 90 countries. On Purple Access, each network's filtering policy sits beside the WiFi it protects. Shield also fails closed and runs across multiple data centres.
Does protective DNS slow browsing down?
It replaces your current resolver rather than adding a hop to the data path, so the question is resolver speed, which you can measure. Filtering can also cut load. Purple's own figures show up to 44% less data and 62% fewer DNS queries on Shield networks. The [latency guide](guide:dns-filtering-latency-speed) shows how to test each SSID before and after the change.
What happens if Shield is unreachable?
Shield fails closed, so an outage does not quietly turn into an unfiltered network. It runs across multiple data centres to prevent that outage, and the Purple platform carries 500 million logins a year at a 99.999% uptime SLA. Keep that design intact by listing only Shield resolvers in option 6. A public secondary resolver turns a fail-closed network into a fail-open one.
How do I migrate to Shield from another DNS filter?
Run the same eight steps, one SSID at a time. Recreate your current category choices as Shield policies first, so behaviour does not change mid-migration. Shorten the lease a day ahead. Then replace the old provider's resolvers in each scope's option 6 and update the firewall rules to allow Shield instead. Pass the verification checks on each SSID before you move to the next.
Does per-SSID protective DNS help with PCI DSS or GDPR?
It supports both, but it does not make you compliant on its own. PCI DSS expects the cardholder environment to be segmented. A payments VLAN with a tight Shield policy adds a DNS control on top of segmentation, which still comes from VLANs and firewall rules. Purple is ISO 27001 certified and GDPR compliant. Shield's query logs give your auditors evidence that the control runs.
Sources
- RFC 2132: DHCP Options and BOOTP Vendor Extensions
- RFC 2131: Dynamic Host Configuration Protocol
- RFC 7858: DNS over Transport Layer Security
- RFC 8484: DNS Queries over HTTPS
- RFC 8106: IPv6 Router Advertisement Options for DNS Configuration
- Microsoft Learn: Set-DhcpServerv4OptionValue
- PCI Security Standards Council
