Download presentation
Presentation is loading. Please wait.
1
Network Layer Protocols
Internet Protocol Version 4 Service Primitives, Header and Functions Internet Protocol Version 6 Internet Control Message Protocol Internet Group Management Protocol Address Resolution Protocol Reverse Address Resolution Protocol Internetwork Packet eXchange (IPX) X.25 Packet Layer Protocol 23/11/10 12-Network
2
Internet Protocol IP is a connectionless network layer protocol
IP is really an internetworking protocol designed to enable datagrams to be forwarded across many types of subnetwork (LANs, MANs and WANs) A key feature of IP is that it is subnetwork independent. It hides the details of the subnetworks from the transport layer 23/11/10 12-Network
3
IP Service Primitives Because IP is connectionless it only has two primitives SEND to transmit a packet DELIVER to receive a packet 23/11/10 12-Network
4
IP Service Primitives SEND primitive Source Address
Destination address Protocol Type of Service indicators Identification Don’t fragment indicator Time to live Data Length Option Data Data DELIVER primitive Source Address Destination address Protocol Type of Service indicators Data Length Option Data Data 23/11/10 12-Network
5
IP v4 Header © Tanenbaum, Prentice Hall International 23/11/10
12-Network
6
IP v4 Header Version = 4 (or 6)
IHL = Internet Header Length (in 32bit words) TOS = Type of Service (3 bit Precedent Field of one bit flags to indicate whether delay, throughput or reliability were most important). Mostly ignored by routers, Now used to indicate class of service for Diffserv protocol) Total Length is of both header and data measured in bytes Identification = Sequence Number same for all fragments, used to reassemble data Fragment Offset indicates where in the original datagram a fragment belongs in 8 byte units DF and MF = Don’t Fragment and More Fragment bits 23/11/10 12-Network
7
IP v4 Header TTL = Time to Live, limits lifetime of a datagram. Each router will decrement TTL by 1 Protocol allows received datagrams to be de-multiplexed to higher layer protocol (TCP/UDP/ICMP) Header Checksum (standard Internet Checksum) calculated with a header checksum field of 0. It has to be recomputed at every router Source and Destination addresses – 32 bit IP addresses Options variable length field up to 40 bytes padded out to a multiple of 4 bytes. One byte option code followed by variable length option. Covers such things as security, source routing, route recording and time-stamping 23/11/10 12-Network
8
IP Functions Fragmentation Encapsulation Addressing Multiplexing
Type of Service Flow Control Error Control 23/11/10 12-Network
9
IP Fragmentation IP will fragment a datagram when its size is greater than the Maximum Transfer Unit size of a subnetwork IP will transmit each fragment as a separate packet with a full IP header All fragments are the same size (the MTU size) except the last one All fragments have the same identification number Each fragment has a different fragment offset which is the byte position in the original datagram measured in units of 8 bytes starting at 0 23/11/10 12-Network
10
Don’t Fragment Bit IP cannot fragment a datagram where the DF bit is set. If a datagram with DF set is too large to transmit over a subnetwork, it must discard it It must also send an ICMP Destination Unreachable packet back to the source 23/11/10 12-Network
11
More Fragments Bit When IP fragments a datagram it sets the MF bit in all the fragments except the last one. IP at the destination knows, when it sees datagrams with the MF bit set, that it has to reassemble fragments up until it receives a fragment with the MF bit not set 23/11/10 12-Network
12
IP Reassembly IP reassembles fragments at the host to avoid datagrams being continually fragmented and reassembled on their path The main disadvantage of this approach is efficiency. Many IP fragments will require a lot of IP protocol header overhead (typically 20 bytes for each fragment) Also the probability of one of many fragments being discarded is higher than one large datagram being discarded. If one fragment is lost, all fragments have to be re-transmitted which will cause delay 23/11/10 12-Network
13
IP Encapsulation © Fitzgerald & Dennis, John Wiley & Sons 23/11/10
12-Network
14
IP Addressing © Tanenbaum, Prentice Hall International 23/11/10
12-Network
15
IP Address Classes (RFC 1117)
Class ID 1st Octet Networks Hosts Purpose A 01 1-126 126 16,777,214 Large Networks 127 Loopback B 10 16,382 65,534 Medium Networks C 110 2,097,152 254 Small Networks D 1110 Multicast E 1111 Experimental 23/11/10 12-Network
16
IP Addressing Network addresses are hierarchically structured into a network portion and a host portion Routers look up the network portion of addresses in routing tables They find the network portion by ANDing the IP address with an address mask, determined by the class of the address 23/11/10 12-Network
17
IP Addressing Subnetting
The IP addressing scheme was found to be too inflexible and wasteful of address space. Also routing tables were becoming too big to manage Subnetting allows bits from the host part of the address to be used to define subnetworks This means that an organisation does not have to register IP addresses for all its subnetworks on a site It only has to register its main IP address for the network, which is the only address that needs to appear in routing tables 23/11/10 12-Network
18
Classless IP Addresses
Problem with classful addresses Class B addresses are too popular and were running out Class C addresses are not popular and were plentiful Solution Classless Internet Domain Routing (CIDR) sometimes called supernetting Allows network addresses to be allocated in variable sized blocks without regard to address classes 23/11/10 12-Network
19
IP Multiplexing IP uses the protocol field to multiplex higher level protocols and determine which process the datagram should be handed to at the receiving end Common protocols defined in RFC 1700 are: ICMP IGMP 6 TCP 17 UDP 23/11/10 12-Network
20
IP Flow Control IP does not perform flow control
If IP can’t cope with traffic it must discard datagrams An ICMP Source Quench message could be sent, but this is rarely used as it just adds to congestion IP relies on higher layers to perform flow control 23/11/10 12-Network
21
IP Error Control IP offers no error control for data. It relies on higher layer protocols to detect and recover from errors in its data field IP does detect errors in its headers by means of the Internet checksum. On detecting an error, it simply discards the datagram 23/11/10 12-Network
22
IP Version 4 Problems IP v4 has a number of problems
It is running out of addresses Routing tables are getting very large It has a complex header structure It isn’t very secure It isn’t good at handling real-time data It doesn’t easily support mobility The answer to all these problems and more is IP v6 23/11/10 12-Network
23
IP Version 6 16 byte addresses gives a total of 3*1038 addresses (7*1023 per m2 over the earth) which can be provider based or geographical Simpler header structure. Only 7 fields. Every header has a next header field which indicates the type of the next header (either an option header or a protocol header such as TCP, UDP or ICMP) Better support for options Improved security Better support for Type of Service and hence real-time applications Anycasting addresses (like multicasting but with delivery to just one of a group of addresses) 23/11/10 12-Network
24
IP Version 6 Header © Tanenbaum, Prentice Hall International 23/11/10
12-Network
25
Internet Control Message Protocol
ICMP is a network layer protocol that sits immediately above IP. The ICMP is encapsulated in IP, so its header comes immediately after the IP header ICMP is used for reporting errors and running diagnostic tests 23/11/10 12-Network
26
Principal ICMP Message Types
© Tanenbaum, Prentice Hall International 23/11/10 12-Network
27
Internet Group Management Protocol
Sender Multicast Router Host Multicasting involves a single send operation that results in copies of the sent data being delivered to many receivers 23/11/10 12-Network
28
Internet Group Management Protocol
IGMP is a network layer protocol used on the Internet for managing multicast groups Multicast routers on a LAN must keep track of which addresses on their LAN belong to which groups. This is done by IGMP. Multicast routers receive one packet with a class D IP address for the group and multicast the packet to all the members of the group on the LAN 23/11/10 12-Network
29
Address Resolution Protocol
ARP is a network layer protocol used on the Internet to map an IP address to a physical address used on a LAN ARP broadcasts an ARP packet on the LAN with the IP address it is trying to resolve All hosts check to see if it is their IP address The host with this IP address responds The sender receives the response and sees which physical address it came from The sender then caches the mapping in its ARP table for future use 23/11/10 12-Network
30
Reverse Address Resolution Protocol
RARP is a network layer protocol used on the Internet by a diskless workstation to discover its IP address from a server on its LAN It broadcasts a RARP packet with its physical address as the source The RARP server responds with the station’s IP address RARP is not used very much these days, as diskless workstations obtain their IP addresses via the BOOTP or Dynamic Host Configuration Protocol that runs above UDP 23/11/10 12-Network
31
Internetwork Packet eXchange (IPX)
IPX is a connectionless network layer protocol, similar to IP, developed by Novell for use with their NetWare product It uses 12 byte addresses which incorporate 6 byte physical addresses, so no mapping is needed between IPX addresses and physical addresses 23/11/10 12-Network
32
X.25 Packet Layer Protocol
X.25 is a protocol suite developed by CCITT (now ITU-T) as a standard for public packet-switched data network. It consists of a physical, data link and network layer. The network layer, called the PLP, is a connection-oriented network that uses 14 digit (7 byte binary coded decimal) addresses X.25 is a legacy corporate network technology, but there is some of it still about. It has a good geographical spread as most PTTs/Telcos run an X.25 network 23/11/10 12-Network
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.