Internet Protocol Addresses

126
 Internet Protocol Addresses
• This chapter
• introduces the address scheme of the Internet Protocol (IPv4 and IPv6)

• describes subnet and classless addressing

• shows how the original IPv4 address scheme was divided into classes

• discusses special IP addresses

• Routing Part II: Aggregation

Computer Networks 127

 Internet Protocol Addresses
• From the last chapters:
• Di erent networks use di erent (layer 2) addressing schemes (an ATM
network does not "understand" an Ethernet MAC address, nor does a
telephone network)

• To connect these networks, however, unique, standardized addresses are

necessary
• -> Introduction of additional, globally standardised addresses for all computers (= hosts)
that communicate via the Internet!

• -> Cross-network Internet address (= layer 3 address)

Computer Networks 128

ff
ff
 RFC 791
• RFC 791 - INTERNET PROTOCOL - September 1981

• The IP standard requires that each host (more precisely: network card) is assigned a
unique number
• the Internet Protocol address (IP address)

• IPv4 (RFC 791): 32-bit number

• Approx. 4 billion addresses (=2 )
32 [9 IP addresses per km 2 of the earth's surface]

• IPv6 (RFC 4291): 128-bit number

• That's not four times as many addresses, that’s four times as many bits

• Approx. 667 quadrillion addresses per mm2 of the earth's surface

• As with any numbering scheme, only a fraction of this is used

• Common: Each host has several IPv6 addresses

• Every packet sent over the Internet contains the IP address of the sender (source) and
the receiver (destination)

Computer Networks 129

 IPv4
• The 32 bit IPv4 address is represented in dotted decimal
notation:

• Each octet (byte) is represented as an unsigned integer in

decimal notation

• The octets are separated by a dot

• Address range 0.0.0.0 to 255.255.255.255

Computer Networks 130

 IPv6
• The 128 bit IPv6 address is represented in colon hexadecimal
notation:

• Two bytes of the address are expressed as a hexadecimal value

(0000 to ) and separated by a colon

• Leading zeros can be omitted: 2001:0:2c4:0:0:a12b:1:abc0

• Case insensitive

Computer Networks 131

ff
ff
 IPv6
• Consecutive zero elds can be abbreviated by :: , but only once per address
• 2001:0000:0234:0000:0000:A12B:0001:ABC0

• 2001:0:234::A12B:1:ABC0

• Not valid: 2001::234::A12B:1:ABC0

• Further examples
• FF02:0:0:0:0:0:0:0:1 -> FF02::1

• 0:0:0:0:0:0:0:1 -> ::1

• 0:0:0:0:0:0:0:0 -> ::

• Side note: In a web browser (URL) the IPv6 address must be given in square brackets:
• http://[2001:0:234::A12B:1:ABC0]:8080/index.html

Computer Networks 132

fi
 IP Addresses
• The IP address is made up of the network ID (pre x) and interface ID (su x)

• Pre x is the address of the physical network

• Su x is the address of the computer within this network

• More precisely: the network card of the computer

• This hierarchy simpli es routing considerably, as routers and their tables

• only take care of the delivery to the correct network.

• they therefore only have to evaluate the pre x and

• do not need to know the addresses of the individual stations within the network

Computer Networks 133

ffi
fi
fi
fi
fi
ffi
 IP Addresses
• Example: Three physical networks, IPv4
• n is 24 everywhere, i.e. 3 byte pre x

• The rst three numbers of the IP address are

the network ID,
• the same for all computers in the same network.

• The last number is the interface ID,

• di erent for computers in the same network

• Router has 3 interfaces, therefore also 3 IP

addresses (one per network)

Computer Networks 134

ff
fi
fi
 IP Addresses
• The IP address hierarchy is conditional on the Internet:
• Each computer (more precisely: interface) is assigned a unique address
worldwide

• The assignment of the network ID must be globally coordinated

• No two networks may have the same network ID

• Su xes can be assigned locally (no global coordination necessary)

• Routers only have to maintain network pre xes, but not individual
addresses in their routing tables

Computer Networks 135

ffi
fi
 Subnet
• All computers (interfaces) with the
same network ID (pre x) form a
subnet

• All computers (interfaces) in a subnet

can reach each other directly, i.e.
without passing through a router

• Example: Three subnets that are

connected to a router. The network ID
is in the rst 24 bits here.
• Subnet mask: /24

Computer Networks 136

fi
fi
 Subnet mask
• Speci es the number n of bits of the network ID. One (IPv6) or
two (IPv4) display options:
inh tuyen lien mien không phan loai
• CIDR representation (Classless Interdomain Routing):
• Is appended to the IP address using /n

• IPv4: 223.1.1.4 /24 -> The left 24 bits of 223.1.1.4 are the network ID

• IPv6: F001::1 /64 -> The left 64 bits of F001::1 are the network ID

• Classic IPv4 representation of /n by 32-bit number:

• whose left n bits are "one" and the remaining bits are "zero"

• 32-bit number is speci ed in dotted-decimal notation: 255.255.255.0

Computer Networks 137

fi
fi
 Subnet mask
• CIDR representation (Classless Interdomain Routing):
• Is appended to the IP address using /n

• IPv4: 223.1.1.4 /24 -> The left 24 bits of 223.1.1.4 are the network ID

• IPv6: F001::1 /64 -> The left 64 bits of F001::1 are the network ID

• Classic IPv4 representation of /n by 32-bit number:

• whose left n bits are "one" and the remaining bits are "zero"
CIDR Classic IPv4
/3 224.0.0.0 (=11100000.00000000.00000000.00000000b)
/24 255.255.255.0 (=11111111.11111111.11111111.00000000b)
/27 255.255.255.224 (=11111111.11111111.11111111.11100000b)

Computer Networks 138

 Subnet
• How many hosts (more precisely: interfaces) t into a subnet with subnet mask /n?

• IPv4: Number of bits for interface ID = (32-n)

• -> 2 (32-n) interface IDs can be formed

• Examples:

• /24 -> 2 32-24 = 28 = 256 Interface IDs -> 256 Hosts

• /25 -> 2 32-25 = 27 = 128 Interface IDs -> 128 Hosts

• /30 -> 2 32-30 = 22 = 4 Interface IDs -> 4 Hosts

• IPv6: Number of bits for interface ID = (128-n)

• /64 -> 2 128-64 = 264 = Interface IDs -> 1.8*10 19 Hosts

• Note: The larger the subnet mask, the fewer IP addresses are available for the
subnet

Computer Networks 139

fi
 Subnet
• IPv4: two su xes are reserved and may not be assigned to a host:
• The smallest interface su x (for /24 e.g. .0, for /8 e.g. .0.0.0) stands for the entire subnet and
represents the network address (e.g. 192.168.1.0/24)

• The largest interface su x represents the broadcast address in the subnet (e.g. 192.168.178.255/24)

• IPv6: the smallest interface su x represents the network address.

• Examples:
• 192.168.1.0/24 -> Network address of the /24 network 192.168.1.0/24

• 10.0.0.0/8 -> Network address of this /8 network

• 192.168.1.255/24 -> Broadcast address of the network 192.168.1.0/24

• 10.255.255. 255/8 -> Broadcast address of the network 10.0.0.0/8

• 2001:1::/64 -> Network address of this /64 network

• 2001:1::1/64 -> Host address 1 of the network 2001:1::/64

Computer Networks 140

ffi
ffi
ffi
ffi
 Subnet
• What is the maximum number of subnets with subnet mask /n?

• Number of bits for network ID = n

• -> 2 n network IDs can be formed

• Example:

• /8 -> 2 =256
8 network IDs -> max. 256 subnets with /8

• /9 -> 2 =512
9 network IDs -> max. 512 subnets with /9

• Note: The larger the subnet mask, the more corresponding subnets there
are

Computer Networks 141

 Subnet
• The pre x requires enough bits so that each physical network
can be assigned a unique NW number

• The su x requires enough bits so that each computer can be

assigned a unique su x

Computer Networks 142

ffi
fi
ffi
 IP Addresses
• The byte-wise former classes of Classfull Addressing:
• A -> /8

• B -> /16

• C -> /24

• 1993: Introduction of CIDR

• Reduction in the size of the global routing table

• No "waste" of IP addresses due to classes that are too large or too small

• In contrast to class-based addressing: Subnet mask can now have any length. (/8, /
9, ... /19, /20, /21 …) -> More ne-grained division of the address space possible

• There are no classes in IPv6

Computer Networks 143

fi
 IP Addresses
• Example: A company has 300 computers in a subnet:

• A Class C /24 network o ers 2 32-24 -2 = 254 addresses -> too

few.

• A Class B /16 network o ers 2 32-16 -2 = 65534 addresses -> far

too many.

• With CIDR: /23 network with 2 32-23 -2 = 510 addresses -> ts

(better)

Computer Networks 144

ff
ff
fi
 Allocation of IP Addresses
• Company or network customer receives a pre x, i.e. a network
number from the Internet Service Provider (ISP)

• The ISP in turn receives network numbers from the responsible

Regional Internet Registry (RIR); (Europe: RIPE NCC)

• There are 5 RIRs worldwide

Computer Networks 145

fi
 Allocation of IP Addresses
• Speci cation of IP address blocks

• Previously: Representation "IP address /n" indicates that the pre x of the IP
address consists of n bits.

• Further use of the "IP address/n" representation as a CIDR address block:

• IP address/n stands for all IP addresses that match the IP address in the rst n bits

• This set of IP addresses is referred to as the IP address/n address block

• Examples
• 10.0.0.0/16 stands for all IP addresses from 10.0.0.0 to 10.0.255.255

• 2000:: /3 stands for the address block 2000:0:0:0:0:0:0:0:0:0 to

3FFF:FFFF:FFFF:FFFF:FFFF:FFFF:FFFF:FFFF:FFFF

• The address block 12.0.0.0 - 12.255.255.255 is called 12.0.0.0/8.

Computer Networks 146

fi
fi
fi
 Allocation of IP Addresses

• IPv6 pre x as aggregated route:

• 2001:600::/23 to RIP NCC

• 2001:638::/32 to DFN

• 2001:638:a01::/48 to OTH Regensburg

• 2001:638:a01:109::/64 NW1-Laboratory

• 2001:638:a01:110::/64 NW2-Laboratory

Computer Networks 147

fi
 Allocation of IP Addresses
• The hierarchical allocation of addresses results in a signi cant
reduction of routing tables on the Internet:

Computer Networks 148

fi
 Allocation of IP Addresses
• Example: A Company wants to set up an IP network. Four
physical networks are to be connected:

Computer Networks 149

 Allocation of IP Addresses
• Estimation of the number of computers results in 350 hosts
• -> Company buys a /23 network (510 addresses) from the ISP

Computer Networks 150

 Allocation of IP Addresses
• The NW administrator must create 4 smaller networks (network
address spaces) from the address space:
• First iteration: division into two networks

• Second iteration (division of the two networks into four networks)

• The four /25 networks are still one large /23 network for the ISP

Computer Networks 151

 Allocation of IP Addresses

Computer Networks 152

 Allocation of IP Addresses
• IPv6
• It works in the same way with IPv6, except that you have to consider 128-
bit numbers

• A subnet is (usually) assigned the subnet mask /64

• An ISP usually receives a /32 address range from the RIR that it can
distribute to end customers

• End customer receives a /48 address range (or only /56)

• End customer can use this to operate up to 2 (64-48) =2 16 = 65536 subnets

Computer Networks 153

 Allocation of IP Addresses
• Company receives a /48 network from the ISP: 2001:A:B::/48

• Assigns /64 networks from this:

Computer Networks 154

 Private IP Addresses
• With IPv4, there are network numbers that can be used for private subnets without an application.

• These addresses must not be directly visible on the global Internet (uniqueness)
• Networks without a connection to the Internet

• Networks that are connected to the Internet via Network Address Translation (NAT)

• Internet routers discard packets that contain private addresses

• RFC 1918: Address Allocation for Private Internets

• 1x Class A: 10.0.0.0/8 (10.0.0.0 - 10.255.255.255)

• 16x Class B: 172.16.0.0/12 (172.16.0.0 - 172.31.255.255)

• 256x Class C: 192.168.0.0/16 (192.168.0.0 - 192.168.255.255)

• There are no private addresses with IPv6

• Global reachability of every IPv6 host

• No more NAT

Computer Networks 155

 Network Address Translation - NAT
• What?
• Replacing the source or destination IP address with another IP address

• Usually also the ports (part of OSI layer 4, explanation later)

• This allows computers with "private", non-routable IP addresses to be connected to the

Internet

• Who?
• The router

• Why?
• Because in IPv4 IP addresses are scarce / all gone

• This is why company-internal computers are typically assigned private IP addresses

• e.g. 192.168.x.y

Computer Networks 156

 Network Address Translation - NAT

192.168.1.1
192.168.1.2

Router

238.125.35.133
192.168.1.3
Replaces internal IP with the external IP address

192.168.1.4

Computer Networks 157

 Network address
• Refers to the network itself, not to the hosts of this network

• IP reserves the host address zero for this

• 128.211.0.0/16 identi es the subnet with the pre x 128.211

• 10.0.0.0/8 identi es the subnet with the pre x 10

• Such addresses are used in routing tables, for example, as routers

do not need to know every host, but only individual networks.

• The NW address should never appear as the destination address

in an IP packet

Computer Networks 158

fi
fi
fi
fi
 Directed broadcast address
• To send a copy of a packet to all hosts
• uses a network "Broadcasting"

• When a packet is sent to a directed broadcast

address
• a single packet travels through the Internet until it reaches
the speci ed network

• then it is sent to all hosts of the network

• Directed broadcast address

• Su x consists of all 1 bits, e.g. 1.2.255.255/16

Computer Networks 159

ffi
fi
 Local broadcast address
• Addresses all hosts of the local subnet

• Is used during startup by a computer that does not yet know its
network address

• IP reserves the address for this, which consists only of 1 bits

(255.255.255.255)

Computer Networks 160

 Dynamic Host Con guration Protocol (DHCP)
• A computer must know its IP address in order to send or receive
packets
• Each packet contains the address of the source and the destination

• The TCP/IP family includes protocols that a computer can use to

automatically obtain its IP address at startup: DHCP
• A newly connected computer sends a local broadcast packet to
255.255.255.255 with the source IP 0.0.0.0

• A DHCP server (e.g. a router) receives the request an assigns an IP address

to the device

Computer Networks 161

fi
 Loopback address
• Each host has a virtual network card, the so-called loopback
interface
• Packets sent to the loopback address are delivered to the host's own
computer

• With IPv4, the pre x 127.0.0.0/8 is reserved for this purpose

• The most popular loopback address 127.0.0.1 is also

referred to as localhost

• Purpose:
• Error diagnosis

• Communication between two network applications on the same

computer
• e.g: Media server that can be accessed both, from the network and from the local
PC via client software

• Other examples are print servers, HTTP servers, etc. ... all can be
contacted via 127.0.0.1 on the local computer

Computer Networks 162

fi
 Special IPv6 addresses
• Unspeci ed: 0:0:0:0:0:0:0:0 or :: (like 0.0.0.0 in IPv4)

• Loopback: 0:0:0:0:0:0:0:1 or ::1 (like 127.0.0.1 in IPv4)

• to test whether your IPv6 stack works: –ping ::1

• Multicast: FF< ags><scope>::<multicast group>

Computer Networks 163

fi
fl
 IPv6 global addresses
n bits m bits 128-n-m bits
Global unicast prefix Subnet ID Interface ID
• 128 bit Total
• >= 48 bit global unicast pre x

• <= 16 bit for subnet

• About 64 bit for interface ID (from MAC address or randomly or assigned

via DHCP)

• Addresses start with

• 2001:

• More precisely: The rst byte has the bits 001x xxxx

Computer Networks 164

fi
fi
 Outlook
• The Internet Protocol (IP) is the common "language" of all
Internet hosts

• Now we know how the addresses work, we can look at further

details

Computer Networks 165