DNS: how a site's name leads to its server

Useful to know

For the person, an address like example.com is a convenient way to remember a site. However, information about where to send the request is required for the network connection. The DNS, domain name system is involved in finding this information. It is a separate layer from the site itself: it can correctly point to the server, and the application on it has a problem. This distinction helps to understand why "the domain works" and "the page opens" are not exactly the same statements. [1]

Typically, the device addresses the question to a recursive resolver. If it does not have an appropriate answer in the cache, look for information along the DNS. hierarchy The root servers and top-level domain servers direct it to the authoritative servers for the respective zone. The authoritative server provides the records for which it is responsible. Therefore, the resolver and the authoritative server have different roles: one finds the answer for the client, and the other supports a certain set of data. [1]

DNS records are not just a list of addresses for web pages. The A record contains a IPv4 address and a AAAA - IPv6 address. CNAME indicates a pseudonym for another name. MX directs e-mail, while TXT stores text information used, for example, in domain checks. NS specifies the name servers. Therefore, when moving a site, it is important to know which records serve its page and which support other services, so as not to unnecessarily change the entire configuration. [2]

Caching saves the reoccurrence of the search. The record has TTL, which determines how long it can be kept in cache according to this setting. A longer time limit increases the possibility of reusing the response, but can delay the visibility of a subsequent change. Reducing TTL in the authoritative zone does not instantly delete old copies that have already been received with the previous term. Therefore, preparation for relocation begins before the change of address itself. [3]

Imagine a study example with two visitors. One made a reference just before the change and received the old record. The other asks a resolver that has not yet been saved, and gets the new one. For a time, they can get to different servers, although they spell the same name. This does not necessarily mean that the editorial has disappeared. The example shows why in diagnostics it matters where the answer comes from, not just what is recorded in the control panel. [1, 3]

It is useful to split the check into specific questions: who is the authoritative server, what record it returns and what the used resolver sees. Separately, it is checked whether the web service of the found address corresponds. This approach avoids the aimless change of settings. DNS solves the problem of names and records; the content, certificate, and logic of the application are evaluated at subsequent levels. Splitting these steps makes even a simple move between two hostings much more understandable.