Polytechnic University1 The internetworking solution of the Internet Prof. Malathi Veeraraghavan Elec. & Comp. Engg. Dept/CATT Polytechnic University [email protected]

1 Polytechnic University

The internetworking solutionof the Internet

Prof. Malathi VeeraraghavanElec. & Comp. Engg. Dept/CATT

Polytechnic [email protected]

What is the internetworking problem:

• how to connect different types of networks


Simplest network – one link

Endpoint Endpoint

Endpoint Endpoint

EndpointEndpoint

Switch Switch

One network – same type of switches – link rates can be different

Single networks

Endpoint

Endpoint

Endpoint

A shared link:often used tocreate a LAN


Endpoint

Endpoint

Switch Switch

Network 1

Endpoint

Endpoint

Switch Switch

Network 2IP router

An internetwork

The Internet approach to internetworking

• Have all endpoints speak the IP (Internet Protocol) in addition to their own network protocols

• For loss-sensitive applications: run TCP, an end-to-end transport protocol, irrespective of whether

– both ends are within the same network– the two ends are on different networks

• IP routers are connectionless packet switches – they forward IP packets from one network to another based on the destination IP

address carried in the IP header and information stored in their routing tables

Network 3


Network 1 Network 2

T1

Inter-T:TCP

Inter-N: IP

A

N1

L1

P1

N1

L3

P3

T1

N2

L4

P4

T2

Inter-N: IP

IP router routerrouter

N1

L2

P2 P3

N1

L1 L2

P1 P2

N2

L5 L6

P5 P6

N2

L4 L5

P4 P5

T2

Inter-T:TCP

Inter-N: IP

A

N2

L6

P6

Switch Switch Switch Switch

Endpoint Endpoint

L3

Protocol stacks in the Internet


Today’s most common networksin the Internet

• Ethernet within enterprises using a combination of– shared-medium Ethernet LANs with hubs, or– with Ethernet switches – which are connectionless

packet switches

• PDH/SONET networks in the MAN and WAN domains– Routers are interconnected by T1, T3, OC3 connections

that are set up through a PDH/SONET circuit-switched network

– PPP, Point-to-Point Protocol, is executed on these circuits


Ethernet frame structure (RFC 894 and 893)

CRCDest.Addr

.

Src.Addr. Data

IPdatagram

ARP req./reply

RARP req./reply

Type

66

2 18

PAD

46-1500

28

0800

0806

8035

Type

Type

Type

28 18

PAD

2

2

2

46-1500

4FOCUS:Ethernetaddresses(also calledMAC addresses)are 6 bytes long


PPP frame structure

7E

flag

1

FF

addr

1

03

ctrl

1 2

protocol

<= 1500

data

2

CRC

7E

flag

1

0021 IP datagram

C021 link control data

8021 network control data

296 if low delay


IP Datagram Format

version(4 bits)

headerlength

Type of Service/TOS(8 bits)

Total Length (in bytes)(16 bits)

Identification (16 bits)flags

(3 bits)Fragment Offset (13 bits)

Source IP address (32 bits)

Destination IP address (32 bits)

Options (if any, <=40 bytes)

DATA

>= five 32-bit words

32-bit word0 31

TTL Time-to-Live(8 bits)

Protocol(8 bits)

Header Checksum (16 bits)

FOCUS


Application

TCP

IP

EthernetDriver

User data

User dataApplicationHeader

Application dataTCP Header

Application dataTCP HeaderIP Header

Application dataTCP HeaderIP HeaderEthernetHeader

EthernetTrailer

IP datagram

TCP segment

Ethernet frame

User-plane interworking -Encapsulation

As data moves down the protocol stack, each protocol adds layer-specific control information.


Layers used in ftp

FTPprogram

TCP

IP

mng.poly.edu

EthernetDriver

EthernetDriver

EthernetDriver

IP

FTPprogram

TCP

IP

photon.poly.edu

EthernetDriver

FTP protocol

TCP protocol

IP protocol IP protocol

Ethernetprotocol

Ethernetprotocol

IP router: dibner-gw


Need Internet address and Network address

Host A

Host B

Switch1

Switch2

Ethernet 1

Host C

Host D

Switch3

Switch4

Ethernet 2

IP router

Internetwork

Host E

Host F

Switch Switch

Ethernet 3

Host A sends a packet to Host C:- Places Host C’s IP address in IP header- To get through Ethernet 1, it needs Ethernet address of IP router’s

interface 1- Switch 1 and Switch 2 forward packets based on destination

Ethernet address of IP router’s interface 1- IP router forwards packet to port 2 to reach Host C (based on IP

level routing data using destination IP address of host C)- IP router needs Ethernet address of Host C to send the packet

through Ethernet 2- Switch 3 and 4 forward packets based on destination Ethernet

address of Host C

1 2

3


Addresses for interfaces

• Both IP addresses and Ethernet addresses are assigned per interface, not per node (router or host).

• An IP router has many interfaces; each interface has an IP address; interfaces that connect the IP router to an Ethernet network also have Ethernet addresses, one per interface

• An Ethernet switch has many interfaces; each has an Ethernet address

• A host typically has only one interface; hence it is assigned one IP address and one Ethernet address if its interface is an Ethernet link


FTP session from host mng to photon

128.238.42.105

dibner-gw.poly.edu

photon.poly.edumng.poly.edu

128.238.32.1128.238.42.1 128.238.32.22

Note that IP router dibner-gw has more than one IP address


Packet sent from mng to IP router dibner-gw

mng.poly.edu

router

dibner-gw.poly.edu

src IP address: 128.238.42.105dst IP address: 128.238.32.22src MAC address: 5:6:7:1:a1:fdst MAC address: 0:0:c:1:a2:e

5:6:7:1:a1:f

0:0:c:1:a2:e

• Host mng consults its IP routing table. This says that to reach destination IP address 128.238.32.22, it needs to send the packet to the IP router because this destination is on a different network

• Hence it sends the packet within its Ethernet network to destination Ethernet (MAC) address 0:0:c:1:a2:e because this is the Ethernet address of the router interface that is connected to mng’s Ethernet network. This destination MAC address allows the Ethernet packet (called frame) to be routed through the first Ethernet network – Ethernet switches determine how to route based on destination MAC address


At the IP router, dibner-gw

• When the packet arrives at the IP router, dibner-gw, it looks up its routing table– For destination IP address 128.238.32.22, the routing table shows which

output port to use.

photon.poly.edu(128.238.32.22)

dibner-gw

128.238.42.1

0:0:c:1:a2:e

dibner-gw

128.238.32.1

0:0:c:1:a2:d


Packet sent from mng to IP router dibner-gw

photon.poly.edu

router

dibner-gw.poly.edu

src IP address: 128.238.42.105dst IP address: 128.238.32.22src MAC address: 0:0:c:1:a2:ddst MAC address: 0:0:5e:3f:4d:2c

0:0:c:1:a2:d

0:0:5e:3f:4d:2c

• IP router, dibner-gw, finds MAC address of photon and adds the IP header and Ethernet header to the packet with the four addresses as shown and sends it.

• The destination MAC address allows for routing through the second Ethernet network; each Ethernet switch that the frame encounters will forward packets based on destination MAC address and its routing table.


Intra-network addresses and inter-network addresses

• Consider example:– mng knows that to reach photon it has to route the

packet to the IP router dibner-gw from its IP-level routing data

– it needs to find the MAC address of the router to get through the first Ethernet

– it does this using ARP (Address Resolution Protocol)

• Same thing when dibner-gw needs to send packet to photon.


ARP (Address Resolution Protocol)

ARP

dibner-gw .poly.edu all stations on thesame Ethernet

128.238.32.22

0:0:5e:3f:4d:2c ARP

ARP

mng.poly.eduall stations on the

same Ethernet

128.238.42.1

0:0:c:1:a2:e ARP


ARP and RARP

• The IP protocol uses 32-bit addresses.• Ethernet networks use 48-bit Ethernet (MAC) addresses• The ARP and RARP protocols perform the translation

between IP addresses and MAC layer addresses.• We will discuss ARP for broadcast LANs, particularly

Ethernet LANs.

RARP

Ethernet MACaddress(48 bit)

ARPIP address(32 bit)


Finding MAC address of an interface whose IP address is available

HOST-BHOST-A

ARP

IP

EthernetDriver

1

EthernetDriver

ARP

2

3

IP

4

(1) HOST-A wants to send an IP datagram to HOST-B.

(2) HOST-A broadcasts an ARP request to all stations on the network: “What is the hardware address of HOST-B?”

(3) HOST-B responds with an ARP Reply which contains its hardware address.

(4) HOST-A transmits the IP datagram to HOST-B.


ARP reply

• The ARP reply is sent by the node whose IP address matches the address sent in the ARP request

• All other nodes receiving the broadcast ARP ignore the request (since their IP addresses do not match the address that is being resolved)


ARP cache

• Clearly, sending an ARP request/reply for each IP datagram is inefficient.

• Each station maintains a cache (ARP Cache) of current entries. The entries expire after 20 minutes.

• Everytime the ARP cache is consulted for a MAC address, the expiry timer is reset in common implementations.) at (incomplete)


Whose addresses does a host store on initialization?

• Go to Control Panel – Network on a Windows PC– Point out that a host needs to have initialized

• host IP address

• gateway IP address (default router interface)

• DNS server IP address


For Internet applications, what type of “address” do you need?

• To begin with, a user obtains “domain name” of a host to which the user wants to connect for a web file download, to send email etc.

• Host needs to find IP address corresponding to domain name – it does this by sending a DNS (Domain Name

Service) query to the DNS Server whose IP address is stored on host (as we just saw)


For the file transfer example from mng to photon

ftpprogram

DomainNameServer

mng.poly.edu gatekeeper.poly.edu

photon.poly.edu

128.238.32.22

• Step 1:– mng knows it wants to execute an ftp to photon.poly.edu– sends DNS query to DNS server gatekeeper.poly.edu– gets IP address corresponding to name photon.poly.edu


What other addresses doesmng need?

• With photon’s IP address, mng consults its routing table to see how it can reach this node.

• Show route print command on PC– point out entry for hosts on the same subnet as

having the gateway address as the IP address of the host interface itself

– point out default entry 255.255.255.255 as having the address of the IP router (gateway)


What if photon was on same Ethernet as mng?

• In this case, the IP address of photon would have matched the entry in the routing table that says photon can be reached through the PC’s Ethernet interface

• Therefore it will try to find the Ethernet address of photon to send the packet via its own network– it does this by issuing an ARP


What happened when photon was on a different Ethernet from mng

• Route lookup showed that IP address (of photon) was on some other network and hence packet had to be sent to the default router (gateway)

• mng then needed Ethernet address of default router. The first time this has to be learned by sending an ARP. Then as long as it is used often (before ARP cache expires), mng knows the Ethernet address of its default router and does not have to issue ARP requests each time


Internet transport protocols

• Loss-sensitive applications: TCP– End-to-end – across networks– To ensure error-free delivery, it executes ARQ retranmission

scheme for error correction– Uses window based flow control– Because IP routers are CL packet switches that do not

implement congestion control, TCP implements an end-to-end congestion control

• Delay-sensitive applications: RTP (Real-time Transfer Protocol)– All packets carry a timestamp allowing receiver to determine

when to play out packets


Summary

• The Internet is an internetwork of networks• Any type of network, CS, CL PS, CO PS can be

part of the Internet• “IP router” is the node that interconnects different

networks– CL packet switch

• Need Internet addresses (i.e., IP addresses) and addresses for each network that packets traverse

• Internetwork transport protocols: TCP and RTP

Documents

Polytechnic University1 The internetworking solution of the Internet Prof. Malathi Veeraraghavan Elec. & Comp. Engg. Dept/CATT Polytechnic University [email protected]