DNS and IP address

Basic components of the internet

What is a domain name?

A domain name is a unique name used to identify a website or e-mail address. Every computer or server on the internet has its own IP address . This is a string of numbers that makes every computer findable on the internet. The Domain Name System (DNS) translates domain names into IP addresses and vice versa.  

Suppose you want to visit a website and have to type a series of digits into your browser ... That doesn't work, because we can’t remember those numbers. We type in the web address. Your computer asks the DNS server what the IP address of that website is. The name server replies with the correct IP address, and your computer connects you to the requested website.  

In concrete terms: you type in dns.be. The recursive server of your Internet Service Provider (Telenet, Proximus, etc.) does not know which IP address corresponds to that domain name. But it does know that it should contact DNS Belgium for .be addresses. So it sends its question to the authoritative server of DNS Belgium and gets the correct IP address in reply. 

There are recursive and authoritative DNS servers. A recursive server is one that does not know the answer itself but knows which authoritative server to look to for the answer. For example, dnsbelgium.be is the authoritative server for .be domain names.

One step further with DNSSEC

If a domain name is protected with DNSSEC (link to DNSSEC page), the DNS data is signed with a digital signature. When a user requests a domain name, the resolver validates the signature. At least if the resolver is set up to validate. 

Name server setup 

The be. zone has 6 anycast clusters. Each cluster has several nodes spread geographically around the world. It is the root zone and root name servers that delegate dns queries to these 6 anycast clusters. Each resolver makes its own decision on which one to query to find its answer. Latency and availability are decisive here.  

Advantages of an anycast cluster: 

  • One (or several) failing node(s) has no major impact on operations. Traffic automatically ends up on the other available nodes in that cluster.
  • You have more chance that the nearest node is going to provide the answer, which benefits overall latency. 

So there are two levels of redundancy for the .be zone: 

  • the resolver having a choice of 6 anycast clusters to send its request to 
  • the anycast operation of each anycast cluster individually.

We have also housed name servers in anycast clusters. Behind each name server record are different servers in physically different locations. This has two advantages:  

  • availability is guaranteed and  
  • the name server is closer to the surfer, making data faster