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(a)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(a)ripe.net
