The Internet Protocol (IP)

The Internet Protocol (IP)
An Introduction

Prof. David Bernstein
James Madison University

Computer Science Department

bernstdh@jmu.edu

Overview

Purpose:
- Provides for the delivery of data through a (set of interconnected) packet-switched network(s)
Specifications:
- IPv4 - RFC791
- IPv6 - RFC2460

Frame Format

Version (4 bits)
Internet Header Length (4 bits)
- The length in 32-bit words
Type of Service (8 bits)
- Precedence (3 bits)
- Desired Frame Delay (1 bit; normal or low)
- Desired Frame Throughput (1 bit; normal or high)
- Desired Frame Reliability (1 bit; normal or high)
- Reserved (2 bits, must be 0)
Total Length (16 bits)
- The length in octets
Identification (16 bits)
- Used to assemble fragments
Flags (3 bits)
- Reserved (1 bit, must be 0)
- Fragmentable? (1 bit; not-allowed or allowed)
- Last? (1 bit; yes or no)
Fragment Offset (13 bits)
- 0-based position of this fragment (measured in 64-bit words)
Time to Live (8 bits)
- The remaining hop count (which is reduced by 1 by each router)
Protocol (8 bits)
- Identifies the protocol that should be used to process the payload (see RFC1700 for options)
Header Checksum (16 bits)
Source Address (32 bits)
Destination Address (32 bits)
Options (variable length)
Payload

Moving through the Stack/Layers

Moving Down:
- The IP datagram becomes the payload of the link layer (e.g., Ethernet) frame
Moving Up:
- The payload of the IP datagram is handed off to the protocol (e.g., TCP, UDP) identified in the protocol portion of the header

IP Addresses

"Classfull Addressing"
images/ip-addresses.gif

IP Addresses (cont.)

A Class C (110...) Example

Assigning/Obtaining an IP Address

Sending from Source to Destination

From A to B (Within a Network)

Sending from Source to Destination (cont.)

From A to E (Between Networks)

Subnetting and Netmasks

Netmasks:
- A dotted-quad that can be ANDed with an address to obtain the network portion of that address (e.g., the netmask for a class C address is is 11111111.11111111.11111111.00000000 or 255.255.255.0)
Creating a Subnet:
1. Extend the network portion of the address n bits to the right (creating \(2^n -2\) possible subnets since subnets of all 0s and all 1s are not allowed)
2. Create the appropriate netmask
An Example:
- Consider 134.126.0.0 (i.e., a class B network)
- Extending the network portion by 3 bits yields \(2^3 - 2\) (i.e., 6) possible subnets
- Each subnet can have \(2^{13} -2\) (i.e., 8190) possible nodes
- The netmask is 11111111 11111111 11100000 00000000 (i.e., 255.255.224.0)

Supernetting

IP Fragmentation and Reassembly

The Issue:
- The link/physical layer probably imposes a limit on the maximum frame size
Fragmentation and Reassembly:
- One frame is divided into multiple fragments
- The fragments are sent to the destination
- The fragments are reassembled at the destination

IP Fragmentation and Reassembly (cont.)

The overall length increases by 60 octets because three additional IP headers are required (each of which is 160 bits or 20 octets). In other words, the original data length was 5140 - 20 = 5120. 5120 octets of data requires 4 fragments, each of which has a header of 20 octets. 5120 + 4*20 = 5200 - 5140 = 60.
The offset is listed in multiples of 8 (e.g., 185 x 8 = 1480 = 1500 - 20)

What's Different About IPv6?

Larger address space (128 bit addresses)
Different packet format
Mandatory use of Internet Protocol Security (IPsec)
Some services provided using different protocols (e.g., the services provided by ARP are provided by the Neighbor Discovery Protocol, stateless configuration is provided using ICMPv6, stateful configuration is provided using DHCPv6)