07

• addressing

• path finding/ routing

• fragmentation

For the Internet these tasks are described in

• Internet Protocol Version 4 (IPv4)

-> old but widely used

• Internet Protocol Version 6 (IPv6)

The Internet Corporation for Assigned Names and Numbers (ICANN) is a non-profit organization developing policies and coordinating central services of the Internet like the Domain Name System (DNS).

The Internet Assigned Numbers Authority (IANA) is responsible for IP address allo cation and port numbers for services.

An IP address consists out of a 32bit unsigned number. For better readability for users the address is partitioned into four bytes. As each byte can hold 256 values (0–255) a typical IP address may be 192.168.10.5.

By using this scheme, without taking reserved addresses and other organizational matters into account, IPv4 can provide 4.294.967.296 addresses.

Just like only working with house numbers in our postal services is not feasible to only use IP addresses for addresses. To achieve something like partitioning into cities and streets IPv4 provides the ability to organize addresses into subnets. These subnets simplify finding the computer with the given IP addresses inside the Internet. Our postal services use tables identify the next distribution center a letter or packet should be transported. This process is called routing and uses a very similar approach. Every device connected by IPv4 has a routing table in which the system or the administrator provides information about which IP subnet is connected to which Network Interface Card (NIC) and which IP addresses or IP subnets should be transported to which next distribution points. These distribution points are in general called routers.

Local Area Networks (LANs) are often Ethernets. Within these Ethernets the Address Resolution Protocol (ARP) is responsible for mapping MAC addresses to IP addresses and vice-versa.

For dividing IP addresses into subnets two schemes are available:

• IPv4 Network Classes

• Classless Inter Domain Routing (CIDR)

The first is not used anymore but is described here for historical purposes. The second is the standard in the Internet today

Class A

• Supports: Very large networks (up to ~16 million hosts per network)

Class B

• Supports: Medium-sized networks (up to ~65,000 hosts per network)

Class C

• Supports: Small networks (up to 254 hosts per network)

Class D (Multicast)

• Used for: Multicasting (one-to-many communication)

Class E (Experimental)

• Reserved for: Experimental purposes, not used publicly

The disadvantage of the class approach is the inflexibility of it. A company requesting an IPv4 subnet will get at least 256 IPv4 addresses event only 10 are required. It is not possible to divide a network class even more. An A class network stays an A class network forever.

CIDR (Classless Inter-Domain Routing)

CIDR replaces old network classes by allowing flexible subnetting. It uses a slash ( / ) notation to show how many bits are for the network part of an IPv4 address.

Example:

• 192.168.0.0/24 → First 24 bits are the network, leaving 8 bits for hosts (256 addresses, minus 2 for network & broadcast).

• 192.168.0.0/25 → Splits the network in half, giving 128 addresses per subnet.

A netmask (like 255.255.255.0) can also show which bits are used for the network.

• 10.0.0.0/8

• 172.16.0.0/12

• 192.168.0.0/16

These addresses will never be used for Internet services and no Internet router will forward these addresses. They are in general used for private networks and hidden behind a router with an routable IPv4 address.

Fragmentation happens when a large IPv4 packet needs to be split into smaller pieces to fit the MTU (Maximum Transmission Unit) of a network. Routers split the packet, and the destination host puts the pieces back together.

The MTU is the maximum size in bytes an individual data link layer frame can transfer en bloc. If an IPv4 packet is received from one data link layer and forwarded to another data link layer with a smaller MTU the IPv4 packet has to be split into multiple IPv4 packets.

eering is when two or more large networks, like Internet Service Providers (ISPs) or hosting companies, connect directly to exchange data (IPv4 packets). They do this without using a middleman or third-party network. This direct connection helps to:

• Speed up data transfer (lower latency)

• Reduce costs (no need to pay another provider to carry the traffic)

• Improve reliability (fewer networks in between)

Virtual Private Networks (VPNs) are a different means to interconnect networks. They create a virtual tunnel through another network and use this tunnel as another data-link layer.

Routing IPv4 packets is done like using an Ethernet data-link layer.

NAT is a method used in networks to translate private IP addresses (used inside a local network) into a public IP address (used on the internet), and vice versa. It allows multiple devices on a private network to share a single public IP address when accessing the internet.

Network Adress Translation

version

the used IP protocol version

header length

the length of the IPv4 header, necessary because the dynamic field of options exists

differentiated services

used for quality of service

total length

the total length of the IPv4 packet including header and payload

identification

uniquely identify a group of fragments

DF

do not fragment bit, used to prevent any fragmentation

MF

more fragments, used to show that more fragments are coming

fragment offset

for ordering fragmented IPv4 packets

TTL time to live

number is counted down on every router, if zero is reached the packet is dropped

protocol

identifying the transport layer protocol

header checksum

checksum just for the header, to identify errors source address

the IPV4 source address

destination address

the IPv4 destination address

options

multiple options for organizing packet management

payload

the data which should be transmitted through IPv4

The Address Resolution Protocol (ARP) provides mappings between IPv4 address and data-link layer addresses, for example when running IPV4 over Ethernet.

The Operating System is providing a mapping table between both address types called ARP table. The entries in this table have an expiration time. If an IPv4 packet has to be transmitted to a locally available IPv4 subnet the operating system will transmit an ARP request as a broadcast asking for the owner of the destination IPv4 address.

The responsible device will answer with an ARP response. The operating system stores this information inside the ARP table. By using this table it is able to map network addresses to the data-link layer address.

The Internet Control Message Protocol (ICMP) is a very important IPv4 protocol. It is used every time the Internet network layer needs to transmit any information.

Typical applications using ICMP are ping and traceroute. T

he first is used to check the availability of a network participant and the round trip time of IPV4 packets to it. The other is used to track the path an IPv4 packet is taking through the network.

The Dynamic Host Configuration Protocol (DHCP) is an important management protocol of IPv4 to automatically configure devices already connected through an Ethernet.

DHCP is a Client/ Server protocol. The server is providing configuration data to requesting clients. Configuration data may be IPV4 addresses, netmasks, routes, default gateway, and DNS servers. The server can provide IPv4 addresses randomly within a IPV4 subnet or based on the Ethernet MAC address of the requesting client