JPNIC Translated Document Source document: JPNIC REPORT 3 Date of the source: November 1993 Date of the last update of this translation: October 28, 1995 This is a translation of a JPNIC document. JPNIC provides this translation for convenience of those who can not read Japanese. But it may contain mis-translations, and is by no means official. One should consult the source document written in Japanese for detail. ----------------------------------------------------------------------- IP address allocation Consumption of IP address IP address is one of the most important resource of the Internet, and NIC is in the charge for allocating the same. However JPNIC is in charge for the continental IP address allocation. IP address is different from domain name and the resource is finite. Within a short period Internet has had a tremendous growth and it has caused the serious problem of exhaustion of IP addresses. A couple of years ago, it was widely recommended to use the class B addresses for reducing the routing control information. Within an organization subnets were utilized and it was recommended to have a single routing information for the external organizations. However, within a 32 bit address space, there are only 16383 class B addresses (Figure 1), and if IP addresses are allocated in this manner then the problem has to be faced where there will be no more class B address left. On the other hand, the space supported by class C network addresses appear to be small for most of the organizations and this can be thought as another factor for the above mentioned problem. Compared to the actual address space necessary for most of the organizations, class B network address space is large and this was another reason for the wide use of class B addresses. In order to cope with this problem, the initiative for allocating class C addresses has been taken. CIDR and route control The technological background lies in CIDR (Classless Inter-Domain Routing) which was proposed in 1992. CIDR eliminates the conventional definition of class based network host and it is based on the idea which corresponds to arbitrary sized networks. In case of CIDR routing control, the netmask also announces at the same time, and it integrates the routing information for a part of the total network, which reduces the load of routing control. If this technology gets widespread then the organizations which are presently connected to the Internet and for whom the class C address is too small and class B address is too large, can have a reasonable size of address space. This will cause an efficient use of address space. However, at present not all of the TCP/IP related products can cope with this technology. For that, the following methods are considered. o To each organization a number of class C addresss which is equivalent to a number that is a power of two will be allocated. By employing this, 16 class C addresses can be represented by removing the first 20 bits from the bit pattern. o Within an organization's terminal networks, this will be treated as the conventional class C or it might be used as a subnet. o Depending on the necessity, the load of routing control will be reduced by aggregating the multiple routing information into a single one from the Internet side. If the above strategies are followed, then by forcing a portion of the router to correspond to the CIDR, instead of a class B address a number of class C addresses could be allotted and this would suppress the increasing overhead of routing control. By advancing this idea further, addresses would be allocable by considering the network topology and the investigation is in progress to enable all kind of routing. As given in figure 2, the routing information of NSFNET is enormously increasing. This has already caused a big overhead, and the prevention of the same has become a big problem. In RFC1466, the scope of "Pacific Rim" is mentioned as, from 202.0.0.0 to 203.255.255.255. This means that if this method is applied with routing aggregation then there is a possibility for reducing routing information. That is, if the first 7 bits are 1100101, then it can be concluded that packets are to be send towards Pacific Rim. For example a network with an address space from 203.2.0.0 to 203.2.255.0 can be considered from outside of Pacific Rim as a single network with net 203.2.0.0 and netmask 255.255.0.0. Although at present CIDR is applied to only class C, theoretically it can also be applied to A and B. However, in that case, all the machines connected to the network should correspond to CIDR. Therefore it is considered to be the next step. Why 32 bits? If the address space consisting of 32 bits is small, then it has to be made larger, is a natural thought. Of course the investigation for that is going on. It appears to be a simple problem like replacing the 3 digits exchange number of the 23 ward of Tokyo to 4 digits. However, compared to the telephones where the only exchanges are centralized on NTT, in case of the Internet not only the routers are scattered but also each computer is basically judging the route. That is, there is a possibility that all the computers would not be able cope with the modification. This kind of large scale modification should have to be handled with necessary technology which would support all sorts of problems (for example transportation of a node) related to address formats. The investigation for this is on progress. In reality it will take some time to put the things into practice and at present there is no other way than to use the 32bit address space. Although it is not adequate to comment like this, CIDR seems to have the meaning of gaining time. Internet and IP address IP address is a must for constructing a network by using TCP/IP technology. Once an IP address is in use, its modification is very difficult. So far, the idea was to obtain a formal address for network operations. For this, if in future there is the necessity for external connections then it won't be necessary to change the address. Not only that, even if the network that requires external connections is an internal part of an organization, formal address was obtained for preventing troubles. More over that, in case of large organizations, there was the idea that it is easy to operate as a subnet of class A or class B, and for this there was a tendency to apply for class B addresses. However, at present there is a strong tendency to relate IP address to the Internet on which the whole world is mutually interconnected, and the idea of thinking it as an address for the connection to the Internet is also wide-spreading. Allocation of IP address The name space of an IP address is fixed and it is totally different from the variable sized address space of domain name. Please consult a professional before applying for an IP address. In particular, if there is a plan for connecting the same to the Internet then please consult a provider. The present allocation is based on an estimate of the next 24 months (2 years) network scale. This is because there is no guarantee that the present IP address structure would be the same after 5 years. Among the class A addresses, addresses from 64.0.0.0 to 127.0.0.0 are not meant for allocation (reserved). For the time being, it can be thought that there is no chance for obtaining a new class A address. IANA (Internet Assigned Numbers Authority) is in charge of allocating class A addresses. Based on an estimate of the next 24 months, the conditions for obtaining a class B address is that there has to be more than 32 subnets, more than 4096 addresses and it has to be proved that it is not possible to construct the network with class C blocks (32 or 64). The application for class B address is screened by the Internet Registry (at present InterNIC). Therefore, the reason for the application has to be written in English. The applications are screened in a very strict manner. There is no guarantee for the allocation of a class B address and if the application appears to be groundless (The estimated number of subnets after 24 months, number of hosts in the subnet and that the network can not be constructed with class C blocks should have to be mentioned clearly), then class C blocks will be allocated. JPNIC forwards the application for class B to the InterNIC but that is only the case where JPNIC decides that the class B allocation is really necessary. As the InterNIC screens the documents by considering the remaining address space, the screening procedure is much more severe than that of JPNIC. Therefore, if the documents does not pass the screening of JPNIC, and the applicant send it directly to InterNIC then there is no hope for getting the allocation. JPNIC allocates only class C addresses. One class C address can manage 254 hosts. However, if the convenience for the operation of subnets is considered then it is difficult to utilize the upper limit of the hosts which could be managed by a class C address. JPNIC divides the estimated number of hosts after 24 months by 100 and the number is expressed as a power of 2, and that is the basis for class C address allocation. However the upper limit of allocation is set to 64 and if that number is crossed then it becomes a subject of discussion. The technology to deal with IP address is changing rapidly. Therefore, the applicant might feel that the correspondence from JPNIC is slow and not kind. However, the procedure for allocating the finite resource to everyone all around the world in a fair way is not a simple task. In the past, the number of application intended for connection to the Internet was rather small, but as the prerequisite for commercial service is to get connected to Internet, the applications with the above mentioned intension have increased. IP address is the most important part for the administration of mutual connectivity to Internet. Because of the growth of the Internet the procedure for handling the addresses is changing. We seek the understanding and cooperation from everyone. (Shin Yoshimura, IIJ)