Recursive resolver
What is Recursive resolver?
A recursive resolver is the DNS server a client actually asks, tracking an answer down through the root, TLD and authoritative servers on the client's behalf, then caching it -- the behavior RFC 1034 and RFC 1035 define. AdminKit's resolver test walks you through finding which recursive resolver a connection is actually using.
By AdminKit team, Netbase JSC · Updated
A recursive resolver is the DNS server a client — a computer, phone, router, or application — actually asks when it looks up a domain. Rather than knowing the answer itself, it tracks the answer down on the client’s behalf by querying the root servers, then the relevant top-level-domain servers, then the domain’s authoritative nameservers, following the referral chain defined by RFC 1034 — also tracked on its IETF Datatracker page — and RFC 1035 until it gets a final answer, which it then caches and returns. ICANN’s own explainer describes the same caching behavior: “Recursive resolvers cache the DNS data they receive from authoritative name servers to speed up the resolution process.” NLnet Labs’ Unbound is one open-source implementation of this role, described by its developers as “a validating, recursive, caching DNS resolver” that also supports DNS-over-TLS and DNS-over-HTTPS for clients. Most users never choose a recursive resolver explicitly: their ISP or router assigns one automatically, often with no filtering or logging transparency, and it can be silently overridden by malware or a compromised router. A recursive resolver that answers queries from anyone on the public internet, rather than only its own network’s clients, is a separate, well-known risk — see open DNS resolver for why that misconfiguration is commonly abused for amplification attacks. Confirming which recursive resolver a device is actually using, as opposed to the one configured, is the first step in diagnosing a DNS hijack or an unexpected forwarding override. A privacy-conscious recursive resolver can also apply QNAME minimization, sending each server in that chain only the minimum label it actually needs. The device making that query is almost never a recursive resolver itself — it’s typically a stub resolver, the lightweight OS-level client that just forwards the lookup here.
AdminKit groups recursive resolver under its DNS cluster — browse the AdminKit glossary index for other related terms.
Frequently asked questions
What does a recursive resolver actually do?
It tracks an answer down on the client's behalf, querying the root servers, then the relevant TLD servers, then the domain's authoritative nameservers, then caches and returns the result.
Who chooses which recursive resolver a device uses?
Usually not the user explicitly -- an ISP or router assigns one automatically, and it can be silently overridden by malware or a compromised router.
What's the risk of a recursive resolver answering queries from anyone on the internet?
It's a well-known misconfiguration, commonly abused for DNS amplification attacks, distinct from a resolver that only serves its own network's clients.
Which RFCs define recursive resolution?
RFC 1034 and RFC 1035.