Is DNS Centralised or Decentralised?
DNS was designed as a distributed system — yet in practice, a handful of providers answer most of the internet's queries. That gap between design and reality matters.
The short answer: both
By design, DNS is decentralised: it's a federated hierarchy in which root servers, top-level-domain registries, and millions of independent authoritative nameservers each answer for their own slice of the namespace, with no single machine holding the whole map. In practice, however, DNS hosting has become heavily concentrated — a small number of large providers serve the authoritative records for an enormous share of popular domains.
So the honest answer is that the protocol is decentralised but the operational reality is increasingly centralised — and it's the operational reality that determines what happens when something breaks.
Why concentration is a problem
DNS is the step before everything else: if name resolution fails, your site is offline for practical purposes even though your servers are healthy. When many domains share one DNS provider, that provider becomes a single point of failure for all of them at once. The pattern has played out repeatedly — most famously in the 2016 DDoS attack on the DNS provider Dyn, which took major sites including Twitter, Spotify, and Reddit offline for hours, and in subsequent outages at other large providers that each removed a visible chunk of the web.
Concentration also raises subtler risks: a single provider is a bigger, more attractive target for attackers; a configuration error propagates to every zone the provider hosts; and censorship or policy decisions at one company can affect a large fraction of the namespace.
What decentralised DNS means
Decentralised DNS pushes the infrastructure that answers queries back out toward the edges of the network. Instead of resolving from a handful of large facilities, queries are answered by a distributed network of nodes, with Anycast routing sending each query to the nearest one. The failure of any node — or any whole region — leaves the rest of the network answering as normal, and a volumetric attack has no single chokepoint to saturate.
Practical resilience follows from the same principle applied at the zone level: serving DNS from a distributed network, signing zones with DNSSEC so answers can't be forged, and using health checks with automatic failover so traffic moves away from failed origins without human intervention.
How Edge DNS works
Edge DNS runs on the same distributed network as the rest of the platform, with unlimited DDoS mitigation included.
Global Anycast
Queries are answered from the nearest of 50+ locations worldwide, with sub-10ms average resolution.
Instant Propagation
Record changes propagate globally in under 5 seconds — no waiting for updates to take effect.
Automatic Failover
Health checks are included, and traffic is routed away from unhealthy origins automatically.
DNSSEC
One-click zone signing protects domains from spoofing and cache poisoning.
DNS resilience in practice
Edge DNS is free to use — up to 50 zones with unlimited records and a million queries a month per zone included, with no per-domain fees. Health checks and geo-routing are included rather than sold as extras, and the DNS Examiner will analyse a zone's security, email deliverability, and nameserver configuration in one click. You can see how it stacks up against Cloudflare, Route 53, and DNSimple on the DNS comparison page.
Whichever provider you use, the lesson of the last decade of DNS outages is the same: treat DNS as critical infrastructure, prefer providers with distributed Anycast networks and DDoS protection, enable DNSSEC, and configure failover before you need it. Our DNS Failover guide walks through the setup.
Put your zones on a distributed network
Fast anycast DNS with DNSSEC, health checks, and failover — free for up to 50 zones.