Protocol Layers

9/19/04

Protocol Layers

-protocol software is designed by layers

-each layer solves a different problem

-one software module for each layer

-the set of modules is called a “protocol stack”

-each layer will have a different header

header 1

header 2

header 3

…

data

-the packet headers will be read by each layer

host1 ------data------ host2

layer4 ….header4…. layer4

layer3 ….header3…. layer3

layer2 ….header2…. layer2

layer1 ….header1…. layer1

-layer N in host1 communicates with layer N in host2 using the header N

Internetworking

-the Internet is a collection of networks that use different networking hardware

+ Ethernet + Token Ring + FDDI

+ RS 232 + SNA + ATM, etc

-it is not possible (economically and logistically) to have a single kind of network hardware. No single technology is the best for all the needs.

-heterogeneity is inevitable

-solution:

Have a layer of software that will make the network look homogeneous. This layer is the IP layer.

-Internet layers:
Application – HTTP, FTP, TELNET, etc

Transport - UDP/TCP

Internet - IP

Network \____ eg: Ethernet, RS232, Token Ring, ATM, etc

Physical /

Host 1	Router1		Router 2
App	Application		Application
Trans	Trans		Trans
Int	Internet		Internet
Net	Net	Net	Net	Net
Phys	Phys	Phys	Phys	Phys

iphdr and data stays the same in host1, router1, router2, host2 hardware hdr (ether hdr and RS232 hdr)

changes depending on the network

Implications:

-there is an RFC that tells you how to send IP packets on Ethernet, on Ethernet, there is one for ip on Token Ring, one for ip on RS232 and for every network hardware known

-RFC – Request For Comments

“internet documentation”
titled by ascending numbers
anybody can write an RFC and it will go to technical review

-the internet layer, transport, layer, and application layer implementation are the same independently of the network hardware

-the internet is formed by

hosts – end points
routers - connect networks
networks – connect hosts and routers

host4 host5

XXX XXX

host1 host2 | |

XXX XXX XXXXXXXXXXXXXXXXXXXX FDDI1

|_______|_______ | |

| host3 | |______ Internet

XXX----- XXX XXX Router2

Router1 | | |

| | |

--------------------

| Token Ring |

--------------------

Networks are also represented by clouds

-a router has a different interface for every network it connects

-a router will have a different IP address for each network interface

-a router can connect networks that have different technologies

Internet Goal:

-to hide heterogeneity to the user applications

Layers:

(5) Application – everything else not mentioned below… HTTP, FTP, Telnet, NFS, finger, Samba, etc
(4) Transport – delivers packets from one application in one computer to another application in another computer (doesn’t need to be reliable, ex UDP). If the protocol offers reliability (TCP) then this layer is also in charge of acknowledgements transmissions etc
(3) Internet – delivers packets across the internet composed of multiple networks (routing, fragmentation, etc) – delivers packets from host to host (IP)
(2) Network Interface – media access format and hardware addressing, translates IP addresses to underlying hardware
(1) Physical – basic hardware

Network Interface and Physical are network hardware dependent

Internet Protocol (IP) - Layer 3

-Defines

Internet addressing
packet format
routing
fragmentation

-IP addressing

abstraction
independent of hardware addressing
used for all communication
32 bits
unique value for each network interface in each host (most of the hosts have one network interface)

A router will have a different IP address fro each interface

IP address

-2 parts

- prefix – identifies network

- suffix – identifies a host

-a global authority assigns the unique prefixes to the networks

-a local administrator assigns a unique suffix to a host

128.10.0.0

– prefix (network part) – given by global authority

- suffix (host part) – given by a local authority

9/21/04

Classes of IP addresses

class A -

0|1 7|8 31|

0| Network | host |

class B –

0|1 15|16 31|

1|0 Net | host |

class C

0|1|2|3 23|24 31|

1|1|0| Net | host |

class D

0|1|2|3| 31|

1|1|1|0| Multicast addresses |

class E

0|1|2|3|4| 31|

1|1|1|1|0| reserved for future use |

Class	#Prefix bits	#Max nets	#Suffix bits	Max hosts
A	7	2^7=128	24	2^34 = 16M
B	14	2^14=16384	16	2^16 = 64K
C	31	2^21=2M	8	256

The maximum number of hosts is given by the class of the address

Examples (Dotted decimal notation):

129.52.6.0

1000 0001 0011 0100 0000 0110 0000 0000

Class B address

192.5.48.3

1100 0000 - class C

10.2.0.3

0000 1010 – class A

The network prefix is used in the network backbone to route packets to the corporations, universities, and ISPs that have that network prefix

(Berkeley-128.210.0.0?)------{INTERNET}--------(Purdue-128.10.0.0)

X(access point)

------(Microsoft-192.5.48.0)

The network prefix is given by a global authority (IANA) and the routers in the internet are configured to send packets for that network access point.

Send IP packet from Berkeley computer (128.210.5.8) to 128.10.5.8

-go through access point to Internet to access point to Purdue network to Purdue computer

The suffix in the IP address will be used to route the packet within the corporation, university, or ISP

INTERNET}----X---(Purdue-128.10.0.0)

| |

[Router-128.10.3.0] [Router-128.10.4.0]

-The subnet mask tells us how many bits correspond to the subnet.

1-net/subnet 0-host

Example:

IP net 128.10.0.0

Subnet mask 255.255.255.0

XXX.XXX|XXX|XXX

Network|sub|host

This subnet tells us that we have 256 subnets and 256 hosts per subnet. This is a typical class D partition. Another assignment could be

Subnet mask 255.255.254.0

128 subnets: [1111 111][0

512 hosts subnet: 0000 0000]

-Subnetting is a decision of the network administrator. They will decide the subnet mask

-The routing within the network (128.10.0.0) will be done according to the subnet number

-The routers and hosts withing the network should be aware of the subnet mask to determine

Obtaining subnet number from an IP address:

ipaddr & subnetmask = subnet number

128.10.5.11 & 255.255.255.0 = 128.10.5.0 – subnet number

Host number:

ipaddr – subnet number

128.10.5.11 – 128.10.5.0 = 0.0.0.11 – host number

Routing Algorithms

Each router has a routing table that tells what is the next hop depending on the subnet/net destination

i1 i2 3.3 5.20

(128.10.10.0)---(R1)---(128.10.3.0)---(R2)------

| | |
| | |

[128.10.10.11] | |

(128.10.5.0) (128.10.6.0)—[6.23]

(R3)

{INTERNET}

Subnet Mask Next hop

128.10.3.0 255.255.255.0 deliver directly using interface

128.10.5.0 255.255.255.0 R2 (128.10.3.2)

128.10.6.0 255.255.255.0 R2 (128.10.3.2)

128.10.10.0 255.255.255.0 deliver directly using interface1 (i1)

Default R2 (128.10.3.2)