JPNIC Translated Document Source document: jpnic/ip-addr-tech.txt Date of the source: November 12, 1993 Date of the last update of this translation: June 27, 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-tranlations, and is by no means official. One should consult the source document written in Japanese for detail. ----------------------------------------------------------------------- Status of this memo: JPNIC This memo was prepared by the JEPG/IP Address Task Force,in cooperation with JPNIC which is carrying out the work of allocating IP addresses, as a reference document for the applicants of IP addresses when they are making their applications. Even the address allocation operations at JPNIC are carried out in accordance with this memo. This memo has been prepared keeping in view the technological trends at the time of its preparation. Since this document will be revised frequently considering the various current situations, the readers are cautioned to refer to the latest version available at the time of making their application. Further, note that this document is being revised at least once every six months. The latest version of this document can be obtained by an anonymous ftp from nic.ad.jp:pub/jepg-ip/ip-adddr-tech.txt or by sending an electronic mail addressed to mail-server@nic.ad.jp. --------------------------------------------------------------- TECHNICAL GUIDE TO THE ACQUISITION OF IP ADDRESSES November 12, 1993 JEPG/IP This document can be copied or distributed freely provided that it is not altered. Also, please send any comments on this document to JEPG/IP Address Task Force by electronic mail addressed to- address@crimson.kek.jp 1. Introduction This document, prepared by JEPG/IP, gives the technical guidelines from the point of view of the applicants for IP addresses in order to clarify the best method of making the application. In the world-wide IP Internet, the method of allocating the IP addresses was greatly modified in 1993. The new method is called CIDR (Classless Inter-Domain Routing), For details of CIDR, the reader is referred to the documents listed at the end of this document. At present, the organizations such as JPNIC that are allocating IP addresses are using allocation methods conforming to CIDR. In more specific terms, the current situation is as follows. i) The range of addresses has been determined for different regions of the world, such as "the addresses 202.0.0.0 to 203.255.255.255 are for the pacific region", as given in RFC1466. A part of these addresses have been assigned for Japan and are being allocated by JPNIC. ii) The method of allocating the addresses conforms to RFC1466. More details are given in Section 4 below. iii) Although it was expected that the routing control technology of Internet (in both Japan and in the international Internet) would conform to CIDR before mid-1993, the implementation is somewhat delayed. Until this is realized, some precautions will have to be taken when connecting to Internet using the IP addresses allocated according to this method. These precautions are described in Section 5 below. Further, in the subsequent future (after about three to five years), the current protocol IPv4 will be replaced by the new protocol IPng. Although what type of protocol is this going to be has not yet been decided, in any case, the handling can be expected to be considerably different from the current IP addresses. 2. General Comments 2-1. IP Addresses are finite resources shared internationally. In applying for these addresses, the application should be made only for those that are actually necessary and applying for addresses as a reserve without any plans for using them should be strictly avoided. 2-2. The IP addresses that are officially allocated by JPNIC, etc., should be used by all the networks participating in the domestic and international IP Internets and by any networks that carry out IP connections of some kind with these networks. In all other networks, it is permissible to use privately allocated IP addresses (that is, those that are not officially allocated by JPNIC) as long as there is no trouble caused to third parties. In such cases, it is necessary to take great care in the management of the routing information of private IP addresses, etc. Of course, even when an organization is not directly participating in an IP Internet, it is possible for it to acquire an official IP address allocation. 2-3. As a rule, IP addresses should be acquired in organizational units, such as universities, companies, etc. Although it is possible to obtain a single IP address by an organization consisting of several universities or companies, it is preferable to have prior consultations with the network projects in which participation is being planned, because such acquisition of IP addresses can cause problems at the time of joining network projects, etc. 3. Future Trends At present, the allocation of IP addresses are being made based on the projected future number of hosts, etc., that are voluntarily quoted by the applicant organizations. Thus, the responsibility for these projections lies with the applicants. Since Internet is an international venture being built up by mutual cooperation between the people of the entire world, it is immoral to acquire large number of IP addresses thinking only of one's own interests. However, future projections are always mere projections, and it is not possible to know exactly what the number of hosts will be in the future. At present, in the allocation method being adopted by Internet, the address allocation is being made corresponding to the relatively more accurate projection of the number of hosts after two years. Therefore, after two years, in most organizations the number of addresses will become insufficient and hence requests for more addresses will have to be made again when their number becomes insufficient. It is also likely that in many cases the growth in the number of hosts will fall short of the current forecast and thus no new addresses will be required by such organizations. 4. Selection of the Class and Number of Addresses The Class and the number of IP addresses (the network address parts) required are selected by forecasting the future number of computers connected to the network and the extent of the distribution of the different departments. The following is the method of allocation described in RFC1466. The point to be noted is that, in the CIDR method, since "a reasonably contiguous block"* of Class C addresses can be considered as a single network address, even a group of Class C addresses can be used without any hindrance when the routing control method of the entire Internet becomes compatible with the CIDR method. The applicants should take particular note of this fact because there are many applicants who misunderstand this aspect and apply for Class B addresses thinking that it will not be possible to get connected to Internet if a Class B address is not acquired now. *Note: It is wasteful for the addresses to be merely contiguous, but should be "powers of 2" with reasonably good boundaries. A block of N (N is a power of 2) contiguous Class C addresses can be considered to have a good boundary when the third octet of the youngest address is perfectly divisible by N. For example, while the two blocks- 192.47.210.0-192.47.211.0 have good boundaries, the four blocks- 192.47.210.0-192.47.211.0 have bad boundaries because 210 is not perfectly divisible by 4. 4-1. Class A addresses As a rule, no new allocations of Class A addresses will be made. If anybody wants the allocation of a new Class A address, they should have sufficient reasons to convince the entire Internet community. 4-2. Class B addresses The standard for the allocation of Class B addresses is that: The applicant has plans to create more than 32 subnets after 24 months (two years), and, in addition, has a need for more than 4096 addresses. In order to acquire the allocation of a Class B address, the applicant will have to provide a technical plan proving that a Class B address is definitely required. In addition, if a Class B address is required for a reason other than the above, although it is possible in principle to acquire a Class B address by sufficiently explaining that reason, it is difficult to acquire such an address in actual practice. 4-3. Class C addresses The Class C addresses are allocated in units of powers of 2. Thus, they are allocated in units of 1, 2, 4, 8, 16, etc. Even the Class C addresses are allocated according to the number of hosts forecast to be used after 24 months (two years). It has been explained in RFC1466 that the number of Class C addresses allocated will be equal to the number of addresses required after 24 months divided by 256 (for example, two Class C addresses will be allocated if 500 host addresses are required). However, there may be some difficulties in this when we consider the actual efficiency of use, for example, it is difficult to operate a network with two Class C addresses if that network expands to 500 hosts after two years. In view of this, as a rough measure, the number obtained by dividing the number of hosts after two years by 60 is taken as the actual number of Class C addresses required. This number is rounded up so as to make it a power of 2. This number 60 is selected as a figure that will make the efficiency of use greater than 25% of the part that can use a Class C address. For example, when there are 100 computers capable of TCP/IP communication connected at present to the network, under the assumption that the number of hosts will increase at an annual rate of 30%, the number of hosts will become 170 after two years. In this case, the measure of Class C addresses required will be four (after rounding up to the nearest power of 2). 4-4. Additional comments The above standards for allocation are used only when addresses are acquired for the first time. The above standards will be applied taking into consideration any IP addresses that have already been acquired. Further, the CIDR method considers a block of contiguous 2- to-the-power-of-n Class C addresses (having a good boundary) as a single network address. For example, while the eight Class C addresses of 200.1.8.0 to 200.1.15.0 are considered as one network address, non-contiguous addresses, or blocks that are not in units of powers of 2, or blocks with no good boundaries such as the addresses from 200.1.10.0 to 200.1.17.0 are not considered as a single network address. At present, since JPNIC is allocating so that this condition is satisfied, the applicants need not worry about this aspect if they are acquiring Class C addresses for the first time. However, care will have to be taken when using a block of Class C addresses acquired in the past or when the addresses were acquired in two installments. When an organization already having some IP addresses is applying for additional addresses, it is possible to return the previously acquired addresses and ask JPNIC for the allocation of a contiguous larger block. In this case, although there will be the burden of having to alter all the previously acquired addresses, there will be the advantages of easier future management and of having a single routing information. In particular, in the case of an organization which had acquired a Class B address but in which the rate of its use was poor, it is preferable to return the Class B address and acquire afresh Class C addresses. In addition, it is preferable in terms of routing control that the IP addresses of the organizations planning to be connected to the same network project are as contiguous as possible. Therefore, when such organizations are applying for addresses around the same time, it is preferable that they file their applications in one collective unit. 5. Connection to Internet As was explained in the introduction, there are some precautions to be taken when connecting to domestic or international Internets using the IP addresses allocated according to the CIDR method. The present routing control method of Internet does not conform to the CIDR method. As a result, when an organization to which are allocated 16 contiguous Class C addresses gets connected as it is to Internet, it will be necessary to send 16 routing information items, which puts excessive load on the routing table management of the entire Internet. In view of this, before actually getting connected to Internet, it is necessary to consult beforehand with the network project to be connected to as to how many routing information items will actually be needed to be sent. However, JEPG/IP is of the opinion that for some time to come, it is not necessary to put any limits on the number of routing information items to be sent by one organization, considering the present capabilities of routers, the bandwidth required for exchanging routing information, and the rate of increase in the number of routing information. This, of course, does not mean that it is correct to blindly send routing information that is not required by Internet. This judgment will naturally be subject to change depending on the changes in the situation in the future. The factor that is most likely to affect the situation in the future is the changes in the method of exchanging routing information. In other words, if the method of exchanging routing information is changed to become compatible with CIDR, the load of routing table management will be reduced greatly. The time of its realization is still not clear and the time its implementation within Japan is not certain. 6. Subnets In a large organization, it is possible to manage with a single address by properly using (see reference document RFC950) the subnet technology. Even subnets are limited resources within a single network, and it is very important to manage them in an effective manner. The readers are requested to refer to the document RFC1219 for the method of appropriately setting the net masks in order to be able to carry out effective management of subnets. In particular, in the case of an organization with three or more levels (for example, the faculty, the departments, and the laboratories in a university), in order to make the management simpler, it is possible to allocate blocks for each subnet for each level. However, this increases the number of wasteful subnets thereby making it necessary to have more addresses than actually necessary. Therefore, in the management of IP addresses, it is strongly recommended to have only two levels within an organization (that is, a top level which centrally manages the subnets, and a level that acquires the allocation of subnets and allocates the host addresses to each of the hosts). 7. Addresses for testing The network address 192.0.2 has been set aside in IP Internet as an address for testing. It is preferable to discard all the packets containing this network address at the time of relaying IP packets. It is recommended that all the manufacturers of routers or work stations set the IP address of all the interfaces of the routers to one of these network addresses (such as 192.0.2.1, etc.,) before shipment from the factory in order to avoid any accidents occurring due to these routers, etc., being connected to the network with the factory shipped IP address setting. * JEPG/IP (Japanese Engineering & Planning Group/IP) is a group that carries out the technical investigations and studies required for the smooth operation and development of IP Internets in Japan and makes the recommendations required to meet this goal. This group was started in December 1991. JEPG/IP sets up individual task forces for different technical problems and these task forces carry out the specialized studies. The Address Space Task Force is the place where technical investigations related to IP addresses are made. Appendix: References on CIDR The basic references for CIDR are RFC1519 and RFC1518. In addition, the allocation method based on CIDR and the schedules for change over to that method are described in detail in RFC1466 and RFC1467, and the reader is recommended to go through these documents. Further, a reference document in Japanese is the document of the IP Meeting '92 (pub/net/address-taskforce/CIDR* of the host ftp.riec.tohoku.ac.jp).