Comment le filtrage DNS réduit la consommation de bande passante réseau

How DNS Filtering Reduces Network Bandwidth Consumption. A Purple WiFi Intelligence Briefing. Introduction and Context. Welcome. If you're managing WiFi infrastructure at scale — whether that's a hotel group, a retail estate, a stadium, or a public-sector campus — you've almost certainly had the conversation about bandwidth. Why is the connection slow during peak hours? Why is the ISP bill climbing when concurrent users haven't changed? Why are guests complaining when your headline throughput looks perfectly adequate on paper? The answer, in a significant proportion of cases, is that a large chunk of your available bandwidth is being consumed by traffic that has nothing to do with your users' actual needs. Advertising networks. Tracking pixels. Telemetry beacons. Malware callbacks. These are silent, persistent consumers of your network capacity, and they operate entirely below the radar of most standard network monitoring tools. Today, I want to walk you through how DNS filtering — specifically, blocking unwanted domains at the DNS resolution layer — addresses this problem directly, reduces unnecessary bandwidth consumption, and delivers measurable ROI for network operators. This isn't theoretical. I'll give you real deployment scenarios, configuration guidance, and the numbers you need to make the case internally. Technical Deep-Dive. Let's start with the fundamentals. When a device connects to your WiFi network and a user opens a browser or an app, that device begins making DNS queries. DNS — the Domain Name System — is essentially the phonebook of the internet. Before any data flows, the device asks a DNS resolver: "What is the IP address for this domain?" Only once it receives an answer does it attempt to connect. Now, here's what most network operators don't realise. On a typical public WiFi network, a substantial proportion of DNS queries are not initiated by the user at all. They are generated automatically by the operating system, by apps running in the background, and by web content loaded alongside the pages users actually want to see. A single page load on a modern news website can trigger DNS queries to thirty, forty, or even sixty distinct domains — the vast majority of which are advertising networks, analytics platforms, and third-party trackers. Research from network telemetry providers consistently shows that between twenty and forty percent of all DNS queries on public WiFi networks resolve to domains associated with advertising, tracking, or telemetry. On networks with a high proportion of Android devices — common in retail and hospitality environments — that figure can be higher still, because Android's background telemetry is particularly aggressive. DNS filtering works by intercepting those queries at the resolver level and returning a null response — or a block page — for any domain on a maintained blocklist. The device receives the response in milliseconds, understands that the domain is unavailable, and moves on. Critically, no TCP connection is established, no TLS handshake occurs, and no data payload is transferred. The bandwidth that would have been consumed by that request simply never flows. This is the core efficiency gain. You are not just blocking content — you are preventing the underlying network transactions from occurring at all. Every blocked DNS query represents a connection that was never made, a payload that was never downloaded, and bandwidth that remains available for legitimate traffic. Let's talk about the categories of traffic you're blocking and the bandwidth implications of each. Advertising networks are the largest single category. Ad serving involves not just the ad creative itself — which can be a multi-megabyte video — but also the bidding infrastructure, the impression tracking, the viewability measurement scripts, and the retargeting pixels. A single ad slot on a page can involve DNS queries to a dozen different domains before a single byte of ad content is served. Blocking these domains at the DNS layer eliminates all of that overhead. Telemetry and diagnostics traffic is the second major category. Operating systems — Windows, macOS, iOS, Android — all send regular telemetry to their respective vendors. This traffic is low-bandwidth per device but cumulative. On a network with five hundred concurrent devices, Windows Update telemetry, Apple diagnostic submissions, and Google Play Services check-ins add up to a meaningful and continuous background load. DNS filtering can suppress this traffic selectively, though operators should be aware of the compliance implications in managed device environments. Malware and botnet command-and-control traffic is the third category. Compromised devices on your network — and on a public WiFi network, you should assume some proportion of connected devices are compromised — will attempt to contact command-and-control servers. These connections are typically low-bandwidth individually but can be high-frequency. More importantly, they represent a security risk that goes beyond bandwidth. DNS filtering against threat intelligence feeds blocks these connections before they can exfiltrate data or receive instructions. Now, let's talk about the architecture of a DNS filtering deployment. There are three primary deployment models. The first is cloud-based DNS filtering, where you redirect your network's DNS traffic to a cloud resolver that applies filtering policies before returning results. This is the lowest-friction deployment model. You change the DNS server address in your DHCP configuration, point it to the filtering provider's resolvers, and you're operational within minutes. The filtering rules are maintained by the provider and updated continuously. This model works well for most venue operators and requires no on-premises hardware changes. The second model is on-premises DNS filtering, where you deploy a filtering appliance or virtual machine within your network that acts as the local DNS resolver. This gives you lower latency — particularly relevant in environments where DNS resolution speed affects user experience — and keeps your DNS query logs within your own infrastructure, which can be important for GDPR compliance and data sovereignty requirements. The trade-off is the operational overhead of maintaining the appliance and keeping blocklists current. The third model is integrated filtering within your WiFi management platform. Platforms like Purple integrate DNS filtering directly into the guest WiFi management layer, allowing you to apply filtering policies per SSID, per user segment, or per time of day. This is the most operationally efficient model for multi-venue operators, because policy management is centralised and consistent across your entire estate. Regardless of deployment model, the key technical components are the same. You need a DNS resolver with blocklist capability, a mechanism for updating blocklists — ideally automated and continuous — and a logging and reporting layer that gives you visibility into what's being blocked and why. On the subject of blocklists: the quality of your blocklist is the single most important variable in the effectiveness of your DNS filtering deployment. A well-maintained blocklist will include advertising and tracking domains, malware and phishing domains, and — depending on your policy requirements — categories like adult content, gambling, or social media. Industry-standard sources include the OISD blocklist, Steven Black's hosts project, and commercial threat intelligence feeds from providers like Cisco Umbrella or Cloudflare Gateway. For enterprise deployments, I'd recommend layering at least two sources: a community-maintained advertising blocklist and a commercial threat intelligence feed. Implementation Recommendations and Pitfalls. Let me give you the practical guidance on deployment, and the failure modes I see most often. The most common mistake is deploying DNS filtering without a baseline measurement. Before you enable filtering, run your network for at least two weeks with DNS query logging enabled. Capture the volume of queries, the top queried domains, and the proportion of traffic going to known advertising and tracking domains. This baseline is your before state, and it's what you'll use to demonstrate ROI after deployment. The second common mistake is using an overly aggressive blocklist without testing. Some community blocklists are extremely broad and will block domains that are legitimate dependencies for services your users need. A blocklist that blocks Google's font CDN, for example, will break the rendering of a significant proportion of websites. Before deploying to production, test your chosen blocklist against a representative sample of the websites and applications your users access. Most enterprise DNS filtering platforms include a dry-run or audit mode for exactly this purpose. The third pitfall is failing to account for DNS over HTTPS, or DoH. Modern browsers — Chrome, Firefox, Edge — increasingly use DoH by default, which means they bypass your local DNS resolver entirely and send encrypted DNS queries directly to a cloud resolver like Cloudflare or Google. If your users' browsers are using DoH, your DNS filtering is invisible to those queries. The solution is to either block DoH providers at the firewall level — forcing devices back to your local resolver — or to deploy a DoH-capable filtering resolver that intercepts and filters encrypted DNS traffic. This is an increasingly important consideration and one that catches many operators off guard. For GDPR compliance, ensure that your DNS query logs are handled in accordance with your data retention policy. DNS logs can contain information about users' browsing behaviour, which constitutes personal data under GDPR. Most enterprise DNS filtering platforms provide configurable log retention periods and anonymisation options. If you're operating a guest WiFi network, your privacy policy should reference DNS filtering and data retention practices. Rapid-Fire Questions and Answers. Let me address the questions I hear most often from network operators. Will DNS filtering slow down my network? No. In fact, it typically reduces latency slightly, because blocked queries receive an immediate null response rather than waiting for a connection to a slow or overloaded ad server. The filtering operation itself adds microseconds, not milliseconds. How much bandwidth can I realistically expect to save? In hospitality environments, we typically see between fifteen and thirty percent reduction in total bandwidth consumption after DNS filtering deployment. In retail environments with high Android device density, that figure can reach thirty-five percent. The variation depends on the user population, the device mix, and the aggressiveness of the blocklist. Does DNS filtering affect the guest experience? When configured correctly, no. Users don't notice that ads aren't loading — they notice that pages load faster. The only exception is if your blocklist is too aggressive and starts blocking legitimate content, which is why baseline testing is essential. Can I apply different filtering policies to different SSIDs? Yes, and you should. Your staff network, your guest network, and any IoT or operational network should have distinct filtering policies. Staff networks may need access to domains that are legitimately blocked on guest networks. IoT networks should have the most restrictive policies of all. Summary and Next Steps. To summarise: DNS filtering is one of the highest-ROI, lowest-disruption interventions available to network operators looking to reduce bandwidth consumption and improve network performance. By blocking advertising, tracking, and malware traffic at the DNS resolution layer, you prevent unnecessary network transactions from occurring at all — freeing capacity for legitimate user traffic, reducing ISP costs, and improving the experience for everyone on the network. The implementation path is straightforward. Establish your baseline, select your deployment model — cloud, on-premises, or integrated platform — choose and test your blocklist, deploy with logging enabled, and measure the outcome against your baseline. For multi-venue operators, the integrated platform model — where DNS filtering is managed alongside your guest WiFi, analytics, and access control — delivers the greatest operational efficiency. Purple's WiFi intelligence platform provides exactly this capability, with per-SSID filtering policies, centralised management across your estate, and the reporting you need to demonstrate ROI to your leadership team. If you're ready to take the next step, the Purple team can walk you through a baseline assessment of your current DNS traffic and give you a realistic projection of the bandwidth savings available at your specific venues. Thank you for listening.