DNS: Turning a Name into an Address

Every machine on the internet is reached by an IP address, a number like 93.184.216.34. Nobody wants to memorize those, so DNS, the Domain Name System, keeps a giant directory mapping names to addresses. When you type a domain name, your computer asks a resolver to look it up. That resolver climbs a chain: it asks a root server which server handles the ending, like .com, then asks that top-level domain server which server handles the specific domain, then asks that authoritative server for the actual IP address. Each answer points to the next place to ask, until the chain bottoms out with a real address.

Climbing that whole chain for every single request would be slow, so resolvers cache the answer. Once a lookup resolves a name, the resolver stores the result for a set amount of time and reuses it on the next request, skipping the hops it already walked. This is why the second visit to a site often feels snappier than the first. If the authoritative server for a domain goes down, any cached answers still work until they expire, but once they expire, lookups for that domain fail until the server comes back or another server takes over.

A domain name is not fused to one machine. It is just a pointer stored in a DNS record, and that record can be changed at any time. If a site moves to a new server with a new IP address, the owner updates the DNS record, and the same name now resolves to the new address. This separation of name from machine is what lets sites change hosting providers, add backup servers, or recover from an outage without forcing every visitor to learn a new name. It connects directly to how the rest of the network survives failures: routing around broken paths, and DNS routing around broken servers, are the same idea at different layers.