Root servers – 101
За да прочетете статията на български, кликнете тук.
How exactly does the Internet work? How we find the exact web page through all data? Maybe alliens are involved? Nope… it’s just humans and their machines of course.
A root name server (also called a DNS root server or a root server for short) is responsible for fundamental functions when it comes to translating domain names into IP addresses: it answers client requests in the domain name system’s root zone (the root zone marks the largest layer in the DNS’ name space). Here, the root name server doesn’t execute the name resolution itself and instead informs the requesting client about which other name server (DNS server) it can obtain further information from regarding the desired IP address.
This is carried out via the so-called root zone file, which is an important element of every DNS root server. The file itself only contains a size of roughly 2 MB. However, it contains all the names and IP addresses of all the top-level domains (TLDs). This data belongs to an important function: the root server relies on this file if it names the name server that contains the necessary details of its request.
But even if they only forward requests, root name servers are indispensable when it comes to name resolution. Without them, the DNS would not be able to function in its current form. A root server works on the domain name system’s root and is to some extent plays the most important role when it comes to registering and naming web addresses.
How root servers work
But just how does a root name server help identify a website’s IP address. In order to understand the root server’s mechanics, it helps to first know a thing or two about the fundamental process of name resolution in the DNS.
In addition to an individual internet address (domain name), every internet service features a unique numeric IP address that’s connected to the domain. The website’s alphanumeric name first has to be translated into the IUP address so that the browser can then present the page.
The process of name resolution
The primary role of Domain Name Systems is to translate domain names into IP addresses (also called ‘forward lookup’). The process of online name resolution creates a hierarchically organized process. However, before the DNS can be assigned to carry out name resolution, the applied system general tries to find the needed IP address within its own data.
The number of stations a request passes through and the order in which it passes depends on many different factors. Factors that can influence this process include the user’s operating system or whether or not UDP or NetBIOs over TCP/IP is used as a protocol. The name resolution itself is always processed the same way in the DNS when it runs through different servers. We’ll show you some of the most important phases that this process goes through when searching for a website’s matching IP address and which role the DNS root server plays here.
When it comes to name resolution, many different name servers can be used. However, root nameservers play an important role within this process.: they depict the highest level instance within the name resolution — in case a domain name cannot be translated into an IP via a local resolver or a provider’s DNS server, the root server then becomes the starting point for locating the IP address. And even if the name resolution is always successful in the previously-mentioned step, the necessary information from the past is collected by the DNS root server and stored. For this reason, it’s important that the server is always able to carry out and support your service.
Root name server overview
In total, there are 13 main DNS root servers, each of which is named with the letters ‘A’ to ‘M’. They all have a IPv4 address and most have an IPv6 address. Managing the root server is ICANN’s responsibility (Internet Corporation for Assigned Names and Numbers). These are, however, operated by different institutions that ensure that data exchange in the root zone always remains correct, available, and secure. In addition to their individual operators, this overview also displays the individual root name servers.
DNS-Root-Servers Letters IPv4 address IPv6 address operator
A 220.127.116.11 2001:503:ba3e::2:30 VeriSign
B 18.104.22.168 2001:478:65::53 USC-ISI
C 22.214.171.124 2001:500:2::c Cogent Communications
D 126.96.36.199 2001:500:2d::d University of Maryland
E 188.8.131.52 NASA
F 184.108.40.206 2001:500:2f::f ISC
G 220.127.116.11 U.S. DoD NIC
H 18.104.22.168 2001:500:1::803f:235 US Army Research Lab
I 22.214.171.124 2001:7FE::53 Autonomica
J 126.96.36.199 2001:503:c27::2:30 VeriSign
K 188.8.131.52 2001:7fd::1 RIPE NCC
L 184.108.40.206 2001:500:3::42 ICANN
M 220.127.116.11 2001:dc3::35 WIDE Project
Each of these root name servers contains an identical copy of the root zone file that may need to be updated from time to time—for example when the TLD responsible for the domain name is changed. Changing the root zone file is a relatively complex process: as soon as an application for an update is registered, this is then checked by the IANA (Internet Assigned Numbers Authority; a division of ICANN). If everything appears to be correct then the US Department of Commerce has to approve of the application given that ICANN is contractually obliged to this entity. Only then is the changed implemented in the root zone by VerisSign, which also operates two root servers, in the root zone.
DNS root server’s security measures
Root servers are confronted with a large number of requests day in, day out. A large number of the 13 root name servers isn’t simply answered by the clients’ request alone; this is done in cooperation with other servers as well. However, there are far more than simply 13 different servers that take care of the root zone requests; all in all, there are hundreds of such scattered throughout the world that are responsible for this task. Most of the servers are located in the United States or Europe.
The fact that these servers are so spread out helps with load balancing and hence increases the reliability of root servers: before Anycast came along, there were only the 13 main root name servers that were able to take care of answering requests. Given that 10 of these are located in the United States, Anycast technology first made this relatively decentralized request processing in the root zone possible. The worldwide distribution of servers furthermore makes for shorter access times when it comes to processing requests, given that the server always answers these in the shortest ways.
A further security measure in terms of the limits of the used root name server’s capacities during normal operation: only a third of the available computing resources are used by servers. This helps ensure that name resolution can still be carried out when multiple DNS root servers experience shortages: in such cases, the rest of the active servers take on the requests that were actually meant to be sent to the downed server.
Following this, various DDoS attacks on DNS root servers didn’t have any success in the past, as their security set-ups were just too strong. Those operating the 13 root servers know only too well what their servers mean for the internet: without them, addressing internet services is no longer feasible.