Dear All, Below is a first draft of the "Hints" supporting documentation. The production of this document is a minuted action item from the last RIPE meeting. The content of the document and the questions below will be discussed at the RIPE meeting next week. Please bring your comments to the meeting. 1. Class D procedure - is the assignment of these within the scope of this procedure? 2. The issue of non-contigous subnets (eg multihomed orgs using a subnetted Class B) and the potential difficulties thereof? do we wish to give advice on this 2. Is there a need for a short Appendix describing how to find a NOC of Last resort (cf App 2 on service providers)? ----------------------cut here---------------------- DRAFT DRAFT DRAFT DRAFT DRAFT DRAFT DRAFT DRAFT DRAFT DRAFT DRAFT DRAFT HINTS FOR ORGANISATIONS REQUESTING IP NETWORK NUMBERS Bob Day, Anne Lord ripe-draft This document is intended to complement and support the information described in the "European IP network number application form and template" (see RIPE document ID: ripe- 83). The aim of the document is to guide you in your choice of class of IP network number so that you choose that which is best suited to your organisations needs. The document is motivated by the large number of applications that are received for Class B address that are not in fact allocated. This accounts for approximately 90% of all class B applications. It is a time consuming and often lengthy process explaining to organisations why their application has been rejected, or why it is taking longer to process, which we hope can be lessened with the publication of this document.  - 2 - Contents 1 Background ................................................ 2 IP network number scarcity ................................ 3 IP addressing ............................................. 3.1 Subnetting ................................................ 4 Choosing the Class of Network Number ...................... 4.1 Using a Single Class C Network Number ..................... 4.2 Using a Block of Class C Network Numbers .................. 4.3 Applying for a Class B Network Number ..................... Appendix 1: Supernetting ........................................ Appendix 2: What to do if you need a Service Provider ...........  - 3 - Copyright c 1993 Whilst every effort has been taken to ensure accuracy, the RIPE NCC does not accept any responsibility for loss or damage arising from the use of information found within this document. Material from this document may be incorporated in other technical documentation, subject to prior agreement from, and acknowledgement of, the RIPE NCC. 1 Background The arrangements for the allocation of Internet (IP) network numbers have recently been revised. Previously these numbers were assigned only by the Network Information Centre (NIC) of the Defense Data Net- work (DDN) in the US. This was done by consensus on behalf of the whole Internet community. Following the change of arrangements, the DDN NIC still has overall responsibility for the allocation of network numbers but it has delegated the actual assignment process on a regional basis. In Europe the delegated authority is the Network Coordination Centre (NCC) run by RIPE under the auspices of RARE. The NCC has further delegated a number of IP ``service providers'' to assign numbers for networks connecting to their respective service networks. The <local example> is one of these service providers (it provides the <local cc: IP Service) and consequently now handles the assignment of "C" network numbers to networks connecting to the <network>. 2 IP network number global scarcity The Internet authorities are increasingly concerned about the possi- bility of exhaustion of the IP address space as a result of the recent explosive growth of the Internet. They have decided upon certain measures to attempt to conserve address space, and other solutions are currently under debate in the community. This is now a matter of some concern. Further detail on the measures decided upon so far is given in Appendix 1 of this document. One of the measures currently practised by the Internet community is to carefully review each and every application for network numbers with respect to its merit on technical grounds. Strict criteria are applied to all organisations, regardless of type, to ensure that the remaining address space is distributed as effectively as possible.  - 4 - 3 IP Addressing The IP address of an end system attached to an IP network is composed of two elements: - the network number identifying to which network the end system is attached (uniquely amongst all the IP networks that constitute the Internet); - the host number identifying the end system on that network. The entire address is a 32 bit quantity. The usual means of represent- ing an address is to write it as a series of four decimal numbers, each representing 8 bits of the entire address, and separated by periods. Thus, for example, the address: 192.100.100.27 would represent the end system numbered ``27'' on the IP network with number ``192.100.100''. It is the requirement for global uniqueness of the network number that leads to the need for co-ordination in the assignment of these numbers. IP network numbers are divided into a number of ``classes'', each of which allows a different maximum number of end systems to be attached to the network it represents (ie gives a different maximum number of possible host addresses). Of these there are two classes that will be relevant to an organisation applying for a network number through the <local sp/nic>. A ``Class C'' network number will allow the attachment of up to 256 end systems, a ``Class B'' network will allow up to 65,636 end systems. (In each case two of the end system numbers are reserved for conventional uses, meaning that the number of host numbers available in practice is 254 or 65,634 respectively.) These figures come about because a Class C network number always occupies the first 24 bits of the full IP address, leaving 8 bits for the host number. This gives the possibility of 256 different host numbers, of which one is reserved as a conventional ``broadcast'' address. A Class B network number only occupies the first 16 bits of the full IP address, leaving 16 bits for the host number. An IP implementation can determine the class of a network number by examining the first two bits. If only the first of these is set - ie the top byte is in the range 128 - 191 - it is a Class B number. If both bits are set (and the next bit is unset) - ie the top byte is in the range 192 - 223 - it is a Class C number.  - 5 - Recently there has been growing interest in the use of Class D numbers as well. These are used to create IP multicast addresses - ie if a system transmits a datagram to an address within a Class D network it will be delivered simulataneously to a group of hosts, rather than to a single host. IP multicasting has applications in the area of coperative working and conferencing, as well as (potentially) in the support of routing protocols. A Class D network number has the top three bits set - ie the top byte has the value 224 or greater. 3.1 Subnetting Associated with each IP address is an ``address mask''. This is a 32 bit quantity that marks, in a bitwise fashion, which bits of the address are to be treated as the network number component and which are to be treated as the host number component. Where a bit is set in the address mask, the corresponding bit of the address is considered to be part of the network number field. Where the bit is unset in the address mask, the corresponding bit is considered to be part of the host number field. For a Class C address the default address mask is 255.255.255.0 (ie the top 24 bits contain the network number). For a Class B address the default address mask is 255.255.0.0. By use of a non-default address mask, it is possible for the administrator of a Class B network number to break it down into a number of Class C ``subnets''. These could then, for example, be assigned one per department in a University, and routers could be used to connect these together. This would allow a site network to be broken down into a set of autonomous networks, whilst the network as a whole appears to the outside world to have a single (Class B) number. As an illustration, assume that an institution has the Class B number 128.100 assigned to it. The administrator could create 256 Class C subnets by specifying a non-default address mask of 255.255.255.0. This would allocate the top 8 bits of the host number field to be an extension of the network number field. Hence the set of Class C numbers 128.100.0 - 128.100.255 would become available. Of these, the first and last in the range should not be used, as they have conventional meanings. This would leave up to 254 Class C numbers for use. In principle subnetting need not be done on an 8-bit boundary eg an address mask of 255.255.240.0 could be used to produce 16 subnets (14 of them useable), each with a 12- bit host field. In practice, however, subnetting is usually confined to an 8-bit boundary.  - 6 - Subnetting is thus a technique of moving the boundary between the host and network number parts of an address. For it to be useful, the IP implementations of all end systems on the network involved must support it. All must also use the same, centrally defined address mask. 4 Choosing the Class of Network Number An organisation that requires more address space than would be provided by a single Class C network number will by default receive a group of Class C numbers instead. This implies that it will need to structure its site network into separate, interconnected Class C networks. The rest of this section goes into detail as to how the decision as to which class of address to apply for should be approached. The aspects to be considered when making this decision are as follows: - the current requirement in terms of the the number of end systems to be connected; - the likely expansion over the next one or two years; - the feasibility or otherwise of routing between networks on site, if multiple Class C networks are to be used. 4.1 Using a Single Class C Network Number If the requirement in terms of end systems to be connected are modest - perhaps a few tens of systems to be connected (max 255 hosts) - a single Class C network number will be sufficient. This is the simplest and most trouble-free, situation. 4.2 Using a Block of Class C Network Numbers If it is likely that there will be a few hundred end systems connected over the next year or two the default choice will be to ask for an assignment of a block of Class C network numbers. These will need to be organised internally as a set of interconnected networks, using an IP router (or routers) as the means of interconnection. A common organisation is for the site's network operator to assign one Class C network per department, and to connect these together via a site ``backbone''. For example, assume that the site has been allocated four Class C network numbers: 192.100.100 - 192.100.103. These could be assigned to three different departments and a backbone, and a sin- gle router used to interconnect them, as shown in Figure 1.  - 7 - 192.100.100 (backbone) ===o==============o===============o============o=== | | | | +---+ +---+ +---+ +---+ Connection | r | | r | | r | | r | --> to service +---+ +---+ +---+ +---+ provider or | | | other ===o======== ===o========= ===o======== 192.100.101 192.100.102 192.100.103 (Dept A) (Dept B) (Dept C) Figure 1: Interconnection of Class C Networks via a Backbone Network Alternatively, the four networks might be connected via a single router, as shown in Figure 2. The choice of interconnection method will be dictated by the conditions on site, but in all cases some form of IP routing equipment will be needed. 192.100.100 (Dept A) +---+ ============================| | | | 192.100.101 (Dept B) | r | ============================| o | | u | Connection 192.100.102 (Dept C) | t |--> to service ============================| e | provider or | r | other 192.100.103 (Dept D) | | ============================| | +---+ Figure 2: Interconnection of Class C Networks via Single Router A consequence of the recent rapid growth of the Internet is that the number of network numbers that have to be configured into regional and international routers has also grown rapidly. This means that these routers' routing tables have also grown to the point where there is concern as to whether they will continue to operate efficiently. To combat this problem the concept of ``supernetting'' is being intro- duced. This is outlined in Appendix 1 (although it is not necessary to understand the concept to apply for a network number). A practical consequence of this move is that a request for multiple Class C net- work numbers will always result in a contiguous block of numbers  - 8 - being assigned, and that the size of the block will always be a power of two (ie 2, 4, 8, 16 or 32 network numbers etc). 4.3 Applying for a Class B Network Number There may be some circumstances where the use of a single Class B network number, rather than a block of Class C numbers is justified. This may be because the number of end systems to be connected is so large that it becomes cumbersome to use a block of Class C numbers. The guideline given by the Internet NIC (in RFC 1366) is that a site network should utilise a Class B number if, based on a 24 month projection, it requires: - more than 32 network numbers (or subnets), AND - it has more than 4096 end systems to connect. The Class B network number could then be subnetted if necessary, according to the site requirements. Site networks that anticipate requiring less than this amount of address space should, under normal circumstances, apply for a block of Class C network numbers. Another potential reason for the use of a Class B network number is that it may be infeasible for the institution to do the IP routing required on its site network if a block of Class C numbers is used. As shown in Figures 1 and 2 above, this will require the installation of routing equipment - either purpose-built routers or end systems equipped with multiple LAN interfaces and IP routing software. This might be impractical in some cases, on the grounds of existing investment in equipment. It might also be impractical in a situation where the site network is multi-protocol and the routers cannot handle all the protocols involved. MAC level bridging might then be required, along with a single network number across the entire network. In making the decision as to whether a Class B number is necessary, note that many purpose-built routers can bridge as well as route (so-called ``brouters''), so it may be possible to route IP whilst bridging other protocols. Note also that the ``supernetting'' development described in Appendix 1 means in theory that the use of IP routers on site can be avoided in the case where a suitable block of Class C network numbers has been assigned. To help the NICs involved determine whether there is a sufficient case for a Class B network number, the organisation is asked on the ``European IP network number  - 9 - application form'' to supply information relating to the number of hosts and the number of subnets, in use now and predicted for one and for two years' time. Besides there being a sufficient number of hosts to address, the NICs must determine that the network cannot be engineered using a number of contiguous class C networks. If the network consists of a large number of physical networks with relatively small numbers of hosts on each, it will be necessary to consider subnetting class C networks. A large number of subnetworks alone is not sufficient justification for allocation of a class B address. The guideline in RFC 1366 will be applied rigorously. The procedure for deciding whether a Class B number can be allocated is first that the <nic/sp> will assess the case and, if it agrees, will recommend to the RIPE NCC that a Class B network number is allocated to the organisation concerned. The RIPE NCC will also review the case briefly and make a decision in consultation with the <nic/sp> and the organisation concerned. Because of this two stage consultation process the application will most likely take longer than normal to be dealt with.  - 10 - Appendix 1 Supernetting One of the perceived problems arising from the rapid growth of the Internet is the consequent growth in the size of the routing tables held in the various regional and international routers. The increased pressure to use multiple Class C network numbers, rather than a single Class B number, in order to economise on the use of the latter class will add to the size of these routing tables. As a way of mitigating this problem it has been decided to use a route aggregation scheme colloquially known as ``supernetting''. (It is also known as CIDR - Classless Inter Domain Routing, and is described in detail in RFC 1338.) The key to the scheme is that where a block of Class C network numbers is assigned to an organisation's network it is done so as a contiguous block of a size that is a power of two. This means that for routing purposes it will then be possible to treat the entire block as a sin- gle network, albeit with a special address mask. (The address mask associated with an IP address is a 32 bit quantity that marks, in a bitwise fashion, which bits of the address are to be treated as the network number component and which are to be treated as the host number component. For a Class C address the default address mask is 255.255.255.0 - ie the top 24 bits contain the network number. For a Class B address the default address mask is 255.255.0.0.) To illustrate this, take as an example the block of four Class C net- work numbers 192.100.100 - 192.100.103. This can be treated as a sin- gle network number 192.100.100 by using an address mask that specifies the network number component to be only 22 bits rather than 24 bits. This is shown in Figure 3. <--------network-------><---host--> +--------+--------+--------+--------+ | 192 | 100 | 100 | | +--------+--------+--------+--------+ address 11111111.11111111.11111100.00000000 mask (ie. 255.255.252.0) Figure 3: Illustration of a Supernetting Address Mask  - 11 - Because the block of network numbers is of size four, and has been assigned to start with a value divisible by four, it is certain that the bottom two bits of the normal 24 bits used for a Class C network number will be zero. Therefore the address mask can be set to make it appear that these two bits are part of the host number component of the address, and consequently that the networks numbered 192.100.101 - 192.100.103 are subnets of the network numbered 192.100.100. Because the block of network numbers is of size four, and has been assigned to start with a value divisible by four, it is certain that the bottom two bits of the normal 24 bits used for a Class C network number will be zero. Therefore the address mask can be set to make it appear that these two bits are part of the host number component of the address, and consequently that the networks numbered 192.100.101 - 192.100.103 are subnets of the network numbered 192.100.100. The technique is called ``supernetting'' because it employs a similar principle to the established technique of ``subnetting''. In the latter case bits from the host number component of an address are made part of the network number component, in effect creating a range of subnets from a single network number. It will work in theory for any size block of network numbers, provided the block is contiguous and the ``power of two'' criterion is satisfied. Supernetting can work in practice only if the IP implementations of all equipment handling it have been modified to understand it. Other- wise the special address mask involved will appear invalid, and the implementation will treat each network number in the block as representing an individual network. Hence if all the routers in a regional network to which the organisation is attached do implement supernetting they will treat the entire block as representing a single network. Consequently, in this example, there would be only one entry in the regional routers' tables rather than four, but IP traffic for any network contained in this block would still be routed correctly to the organisation concerned. Depending on implementation of supernetting by the major router ven- dors, it is expected that regional and international routers will adopt this scheme in near future. Follow the recommendations of the provider involved. If all end systems on the network of a connecting organisation, and the router used to connect to the outside world implement supernet- ting it will be possible to construct the network using a block of Class C numbers and without the need for router(s) internal to the  - 12 - network. However, it seems very unlikely that this will be the case in the immediate future, and it is best to assume that traditional routing techniques will be required within the site.  - 13 - Appendix 2 What to do if you need a Service Provider If your organisation is planning to connect to the Internet in the near future, then it is recommended that you do this via an IP Service Provider. If you are unsure who your service provider would naturally be, then you can fax or telephone the RIPE NCC who will send details of your connectivity requirements to a mailing list maintained for this purpose. Please supply your contact information which individual IP providers who have subscribed to the list can use to contact you. If you are sending a fax, please mark it: For the attention of : ip-provs@ripe.net We will then transcribe your details to our electronic mailing list. Note that this is the extent of the NCC involvement - it is a matter for individual service providers to decide whether to follow up such a request. RIPE Network Coordination Centre tel: +31 20 592 5065 Kruislaan 409 fax: +31 20 592 5090 1098 SJ Amsterdam email: hostmaster@ripe.net