Protocols & standards

DNS resolver

A server that receives DNS queries from clients and retrieves the answers by querying authoritative servers.
← Back to glossary

What is a DNS resolver?

A DNS resolver is software that obtains DNS answers for an application. In everyday usage, the term often means a recursive resolver: a service that checks its cache and, when necessary, follows DNS referrals to find an answer. Resolvers retrieve more than IP addresses, including mail routing and other record types. RFC 9499 defines the terminology.

Stub, recursive and forwarding roles

A stub resolver usually runs on a device or within an application. It sends queries to a configured recursive service and relies on that service to complete the lookup. A recursive resolver can follow referrals through the DNS hierarchy. A forwarding resolver passes queries to another resolver instead of doing that work itself. One system can combine roles.

Application -> stub resolver -> recursive resolver

|

+-> cached answer, if usable

+-> DNS hierarchy, if needed

A home router may forward queries, and a browser may use its own resolver configuration. Changing an operating system setting therefore does not necessarily change the path used by every application.

What happens on a cache miss?

For a public name, a recursive resolver can obtain referrals from root and TLD servers before reaching the authoritative nameservers. It reuses cached delegation information where possible, so every query does not start at the root. Aliases and other dependencies may require additional lookups. The basic process is described in RFC 1034.

The resolver can cache answers for their TTL and return the remaining lifetime to later clients. It can also cache negative results. NXDOMAIN means the queried name does not exist; an answer with no records of the requested type has a different meaning. RFC 2308 describes negative caching. A timeout or SERVFAIL should not be treated as proof that a name is absent.

Query minimization

Query name minimization reduces the name information sent to servers that only need to provide a referral. For example, a root server can help locate the .com servers without receiving the full name www.example.com. The resolver discloses more labels as needed to proceed.

The current specification is RFC 9156, which obsoletes RFC 7816. Minimization reduces disclosure to parts of the hierarchy; it does not hide the full query from the recursive resolver chosen by the client.

DoH and DoT: transport privacy

DNS over HTTPS and DNS over TLS protect DNS traffic on the encrypted connection between a client and its resolver. Correct authentication and configuration matter. Encryption does not make a resolver anonymous or prevent its operator from seeing the queries it processes.

These transports also do not automatically encrypt every subsequent query from that resolver to authoritative servers, or conceal all browsing activity from every observer. Consider the resolver's published privacy practices and any fallback behavior as well as the transport name.

DNSSEC and filtering

A validating resolver checks DNSSEC signatures where a usable chain of trust exists. It can reject bogus signed data, often returning SERVFAIL. Unsigned data is not automatically bogus. DNSSEC authenticates DNS data; it does not encrypt queries or certify that a website is trustworthy. See the DNSSEC introduction.

Filtering adds another policy layer. It can block selected names or alter responses, which may help meet a security or organizational requirement but can also block a legitimate destination. When comparing resolvers, check validation, filtering, privacy, reliability and support for private network names. During troubleshooting, record the resolver and response status before switching services so you can explain the difference.

Put this knowledge to work

Use DomScan's API to check domain availability, health, and more.