DNS Fundamentals

1Introduction

This lecture explains how the DNSDomain Name System manages domain-name information through a hierarchical, distributed system. It examines the roles of the servers involved in resolution, resource records, caching, and the transports used by DNS.

2A Hierarchical, Distributed Namespace

A domain nameDomain name, such as www.example.com, consists of hierarchically joined labels. The DNS namespace has a root at its top, followed by top-level domainsTop-level domain; TLD such as com and lower domains such as example.com.

DNS does not store all information on one server. It is a distributed system that delegates management of parts of the namespace to multiple organizations and servers.

3Resource Records

The units of information stored by DNS are called resource recordsResource record. Common record types include the following.

DNS therefore does more than convert domain names to IP addresses. Address retrieval through an A record or AAAA record is one important use.

4Resolvers and Authoritative Name Servers

A stub resolverStub resolver accepts DNS queries from applications. It normally sends a query to a recursive resolverRecursive resolver. If the recursive resolver lacks the required answer and delegation information in its cache, it conceptually follows delegations from root name servers through TLD name servers to an authoritative name serverAuthoritative name server. Cached delegation information can allow the process to begin at an intermediate stage.

An authoritative name server supplies the official resource records for a delegated portion of the namespace. A recursive resolver performs iterative queries to multiple name servers on behalf of a client and returns the final answer to the client.

5Resolution and Caching

Retrieving an address for www.example.com conceptually proceeds as follows.

1. An application requests a query through its stub resolver.
2. The stub resolver queries a recursive resolver.
3. If the recursive resolver has a valid cached answer, it returns that answer.
4. If the required information is not cached, it follows delegations involving root, TLD, and authoritative name servers to obtain referrals and an answer. Cached delegation information can allow it to begin partway through this process.
5. The recursive resolver caches the answer and returns it to the stub resolver.

A resource record has a TTLTime to Live. A cache reuses an answer while its TTL remains valid, reducing query latency and load on DNS servers. Conversely, after a record changes, old information can remain in use until existing cached entries expire.

6DNS Transport

Traditional DNS communication normally uses UDP or TCP on port 53. UDP is widely used for ordinary queries, while TCP is used when a UDP response is truncated or an operation requires TCP. DoTDNS over TLS and DoHDNS over HTTPS are alternative mechanisms that encrypt DNS communication.

DNS queries themselves operate over IP networks. DNS must therefore not be interpreted as a separate process that occurs before TCP/IP.

7From Resolution to Web Communication

When a URL host is written as a domain name, a client uses DNS to obtain one or more IP addresses and selects a destination. It then establishes a TCP connection for HTTP/1.1 or HTTP/2. HTTP/3 instead uses QUICQUIC over UDP. When an IP address is specified directly, domain-name resolution may not be required.

8Key Points

• DNS is a hierarchical, distributed namespace and resource-record query system.
• A recursive resolver obtains an answer from its cache or through queries involving authoritative name servers.
• Caches reuse answers during their TTLs to reduce latency and load.
• DNS communication operates over IP using UDP, TCP, or encrypted alternatives.

9Next Lecture