All computers in the Internet (also called hosts) are identified by an unambigous number, the so-called IP address. It has the form xxx.xxx.xxx.xxx, where xxx is a placeholder for numbers between 0 and 255. A typical IP address would be 190.136.66.12. If a host wants to communicate with another host over the Internet, then it sends "Internet packets", which contain an IP number as "address". Because users can remember names better than numbers, the domain name system was introduced. Each domain name corresponds to one IP address. This way the hosts can be accessed via IP addresses and domain names. Due to the unambiguousness requirement of IP addresses, every domain name must also be unique worldwide.

Domain names are organized hierarchically from right to left. The rightmost part after the last dot is called top level domain, the one to its left second level domain or simply domain. All further parts to the left are sub-domains. Domains below the top level domain DE are managed by DENIC eG.

Back to Top

The top level domains (TLDs) are the highest level in the international domain name system. They are the rightmost part of a domain name, right after the last dot. There is a subdivision into general or generic TLDs (gTLDs) like .com, .net and .org and country-specific TLDs (the ccTLDs for country code TLDs) like .de for Germany or .ch for Switzerland. The top level domains .mil, .int, .edu and .gov are also gTLDs, but are reserved for specific user groups like the American government or international organisations. The country codes are managed by the Network Information Centers (NICs), for Germany this would be DENIC eG. The international organization ICANN is responsible for the gTLDs, their operative administration is in the hands of the company Network Solutions.

Back to Top

DENIC eG was, until April 30, 1996, the "Registry of Last Resort" for Germany and did in this function provide unique IP addresses.

Due to the strong growth of the Internet and the technical problems arising from this growth, this procedure is not practicable any longer. Therefore, since April 30, 1996, we do not provide IP addresses anymore.

If you need IP addresses for Internet access, then you can order them from your Internet Service Provider. You can use one of our members or choose from a large list of Local Internet Registries at RIPE.

If you do not want to connect yourself to the Internet, you can probably get by with a so-called private address space. This concept is described in detail in RFC1918 (Address Allocation for Private Internets).

Back to Top

In the Internet every host (computer) has a unique ID, the IP-address, which consists of four bytes. These four bytes are written in decimal notation, seperated by dots.

The administrator of a LAN (Local Area Network) receives from his Network Information Center (NIC) a block of consecutive IP-addresses and assigns them to the connected hosts. This block is called an "Address Range". The number of addresses in an "Address Range" is an exponent of 2. If needed different sized blocks are combined to a larger block.

The following table shows how many addresses can be assigned to hosts with a specific block size (classfull), how many identical bytes the addresses have (prefix) and what subnet mask identifies the hosts.

hosts	prefix	classfull	subnet mask
2	31	-	255 255 255 254
4	30	-	255 255 255 252
8	29	-	255 255 255 248
16	28	-	255 255 255 240
32	27	-	255 255 255 224
64	26	-	255 255 255 192
128	25	-	255 255 255 128
256	24	1C	255 255 255 0
512	23	2C	255 255 254 0
1.024	22	4C	255 255 252 0
2.048	21	8C	255 255 248 0
4.096	20	16C	255 255 240 0
8.192	19	32C	255 255 224 0
16.384	18	64C	255 255 192 0
32.768	17	128C	255 255 128 0
65.536	16	1B	255 255 0 0
131.072	15	2B	255 254 0 0
262.144	14	4B	255 252 0 0
524.288	13	8B	255 248 0 0
1.048.576	12	16B	255 240 0 0
2.097.152	11	32B	255 224 0 0
4.194.304	10	64B	255 192 0 0
8.388.608	9	128B	255 128 0 0
16.777.216	8	1A	255 0 0 0
33.554.432	7	2A	254 0 0 0
67.108.864	6	4A	252 0 0 0
134.217.728	5	8A	248 0 0 0
268.435.456	4	16A	240 0 0 0
536.870.912	3	32A	224 0 0 0
1.073.741.824	2	64A	192 0 0 0
2.147.483.648	1	128A	128 0 0 0
4.294.967.296	0	256A	0 0 0 0

The historical division into class A, class B and class C nets corresponds to "Address Ranges" with a fixed number of addresses. There were only three possible block sizes, which were supposed to correspond to the size of the respective organization. This division has, however, proved to be unflexible and was discontinued in 1994 in favor of the "Address Ranges".

• Class A Address
  It is of the form A.x.x.x with 0 <= A <= 126. Thus there are exactly 127 class A addresses. The address 10.x.x.x is not assigned.

• Class B Address
  It is of the form B1.B2.x.x with 128 <= B1 <= 191 and 0 <= B2 <= 255. With a class B net it is possible to assign addresses to about 65000 hosts. Currently over 61% of all class B addresses have been assigned, so that due to the rapid growth of the Internet a scarcity develops. Therefore class B addresses are only assigned in cases, where the need for a class B net is well-founded and reasonable. The addresses 172.16.0.0 up to 172.31.255.255 are not assigned.

• Class-C-Address
  It is of the form C1.C2.C3.x with 192 <= C1 <= 223 and 0 <= C2, C3 <= 255 and allows up to 255 hosts in one address block. Only 36% of all available class C nets are assigned, so usually, instead of a class B net a number of consecutive class C nets is assigned. The reserved addresses in this range are 192.168.0.0 up to 192.168.255.255.

Back to Top

There are two ways to secure the delegation of a domain:

1. Our primary nameserver delegates the domain to a subordinate nameserver. This means that the addresses of two nameservers are entered into our database. Every request, that reaches our nameserver, is forwarded to this nameserver with the expectation that this nameserver provides additional data for the domain in question. Therefore nameservers are tested for minimal functionality prior to the first time they are entered into our database.

2. Alternatively, we can configure our primary nameserver in such a way, that up to five services like www.exampledomain.de or mail.exampledomain.de are connected with IP addresses of servers, that operate these services. These domains have previously been called MX-ONLY-Domains and are today known as NSentry domains.

Back to Top

DENIC eG in Frankfurt/Main administers the top level domain .de and operates the primary nameservers for the DE-domains. Therefore DENIC eG, which is borne by the German Internet Service Providers (ISPs), is the central registry for all DE-domains.

DENIC eG offers no IP services (Internet access, webspace, etc.) to third parties and also does not manage websites. These services are provided by ISPs.

Back to Top

The only way to carry out a domain registration is via a member of DENIC eG. According to 3 of our statute you can become a member yourself if you are an Internet Service Provider offering all IP services to your customers (E-mail, FTP, WWW, Telnet etc.), if you have two independent uplinks to the Internet and if you possess your own networks (IP address blocks).

If you want to find out more, please write to mitgliedschaft@denic.de and include your postal address or contact our office. We will then send you a membership application form. The financial aspects of becoming a member of DENIC eG: you have to pay an admission fee of 1500 Euro, buy a stake in our cooperative for 1500 Euro and deposit 15000 Euro as a bank guarantee or a time money account pledged to DENIC eG.

We ask for your understanding that we can only tell you about the member prices for domain registration, administration and provider changes if the DENIC board has agreed to your membership application.

Back to Top

Some (unfortunately rudimentary) statistics about the growth of the Internet can be found on our website. This presentation is based on numbers, that are determined by RIPE. More information, e.g. about the method with which these numbers are retrieved, are available under http://www.ripe.net/ripencc/pub-services/stats/hostcount/.

These numbers, however, are not accurate numbers, they should be understood as orders of magnitude. For example, hosts behind a firewall are not reachable and therefore cannot be included in the count. The number of Internet users is also largely unknown. One could multiply the number of hosts with the average number of users per host (perhaps 3-10) to get a roundabout number based on an estimate.

Additionally, there is the question which kinds of Internet connections should be counted:

• There are users and hosts behind firewalls, which consequently are invisible.
• Do we want to count T-Online, AOL- or CompuServe users?
• Does the technical possibility of Internet access count or only the actual access?
• What about dial-up users?
• ...

Unfortunately we do not have more information available, or, respectively, more information does not exist. If you know a publicy accessible source to more accurate data, we kindly ask you to notify us.

Back to Top