Computer Networks and The Internet

Download as pdf or txt
Download as pdf or txt
You are on page 1of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet


air

Comcast
network

AT&T
network

core network

cable

fiber

enterprise network

laptops, tablets, desktops

smartphones

access point

cell tower

1 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

When we talk about computer networks, we are really talking about the Internet
because no networks today are isolated, they are all interconnected forming part of the
Internet

The Internet is so hugh, you can view it as a network of networks and these networks
cooperate with each other by running the same or compatible protocols

It should be noted that the web is not a network but rather just one of many
distributed applications that use the communication services provided by the Internets

Roughly, the Internet operates at two tiers, namely, the access networks and the core
network

Access networks connect users to the Internet and the core network interconnects the
access networks

Access networks are owned and operated by regional ISPs (Internet Service Providers)
such as AT&T and Comcast in Memphis, other ISPs include Pacific Bells, Road
Runner, Cox, Charter, etc. in other states
Memphis was served by Road Runner until it swapped territories with Comcast
Comcast is consistently rated the worst customer service in the country

AT&T owns the twisted copper wires to the homes because of the legacy telephone
network
2 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

These twisted copper wires terminate at the CO (Central Office)

AT&T provides subscribers with DSL (Digital Subscriber Line) gateways that connect
users computers and smartphones to its access network
DSL uses equipment at the CO to condition the signals on the copper wires making it
suitable for carrying digital signals between homes and the CO

On the user side, these gateways support both Ethernet and wifi interfaces, increasing
wifi is the interface of choice

Depending on the distance from user homes to the CO, the maximum speeds of DSL
are around a few Mbps (one million bits per second); the shorter the length of the
copper wire, the faster the DSL will be

If users connect their computers via Ethernet cables, Ethernet frames are directly
exchanged between the user computers and the gateways

If users connect their computers via wifi, Ethernet frames are first encapsulated in wifi
frames and sent to the gateways and the gateways recover the Ethernet frames from the
wifi frames

Between the AT&T gateways and the access network, the Ethernet frames are carried
in ATM (Asynchronous Transfer Mode) cells; ATM cells are fixed size packets of 53
bytes long
3 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

ATM was originated from the telcos for extending the their voice network to a packet
switch network
Because delay and delay variation are important to voice, to control the delay
variation, ATM cells are short fixed at 53 bytes
To control the delay, the path within the ATM network is pre-established reserving
resource in the ATM switches along the path. It uses the so called signalling to find the
best path. Signalling supports routing and resource reservation

AT&T is now building out fiber-to-the-curb (FTTC) using DSLAMs (Digital


Subscriber Loop Access Multiplexers) to terminate both the fiber and twisted copper
pairs from the homes

Because the short distance between the homes and DSLAMs (typically less than half a
mile), the speed can be up to hundreds of Mbps and that is the reason AT&T can offer
triple play (voice, Internet, and TV) service - by the way, the service is called U-verse

Many new development areas around Memphis are still waiting for the U-verse service

Comcast owns the coaxial cable to the homes because of the legacy cable TV network

Coaxial cables inherently support high bandwidth and therefore, high bit rate

Comcast provides subscribers with cable modems that connect users computers and
smartphones to its access network
4 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

Like DSL gateways, cable modems support both Ethernet and wifi interfaces

All cable provider networks are actually Ethernet networks, i.e., based on the physical
layer of CSMA-CD (Carrier Sense Multiple Access - Collision Detection)

Because it is Ethernet, the claimed high speeds (> 100 Mbps) are shared by all
subscribers who share the cable of the neighborhood

Comcast also offers triple play service but generally less reliable than DSL because
telephones were designed for live line, not cable

DSL gateways and cable modems serve two additional important functions, namely,
supplying an IP address to each of the home user devices using DHCP (Dynamic Host
Control Protocol) and translating addresses between the home user IP addresses and
the IP address seen in the Internet

The core network itself is made up of many smaller administratively separate peer core
networks owned and operated by different companies of different countries

The operators of the smaller core networks do not charge each other for the traffic
flowing through their networks

The operators are free to derive and retain revenue from their own customers

The core networks are long-haul, connects no end users, and utilizes expensive fast
core routers interconnected by fibers
5 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

The core network also uses optical switches support DWDM (Dense Wavelength
Division Multiplexing) and wavelength switching

To interoperate between these smaller core networks, the operators agree to run the
same protocols for exchanging traffic between these networks and this protocol is
called the Border Gateway Protocol (BGP)

ISPs have to pay usage fees to the core network operators

The entire Internet is a packet switched network in which it ships packets of user
information from points to points around the world

What is a packet?

The unit of information exchange is called a packet. A packet is a sequence of bits


representing user data, the address of the sender, the address of the recipient, and a
whole lot of other information such as specifying the priority of the data when being
transported inside the network, data protection to ensure the integrity of the packet
when it arrives at the recipient address, dynamic information for keeping track of the
number of hops the packet has traveled so far, etc. A generic packet is shown below.
destination
address field
48 bits

source
address field
48 bits

priority
field
3 bits

data
field
up to 12,000 bits

error protection
field
32 bits

6 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet


What is a packet switched network?

A packet switched network is a collection of packet switches interconnected by


physical links, used for moving packets from points to points
The figure below depicts a packet switched network of seven packet switches
interconnected by physical links in the pattern shown

Some of the switches connect the hosts, at the edge of the network, allowing them to
send and receive packets and other switches do not connect any hosts but serving
internal switching function only

The location of the switches can be localized within a small geographic area like a city
or spread across continents

What is a packet switch?

The figure below shows a 10x10 switch drawing and a rack mount router switch

A packet switch is an electronic device that has two or more input lines for receiving
packets and two or more output lines for transmitting packets

It uses the input packets recipient addresses and directs the packets to the outputs that
lead to the eventual delivery of the packets to the intended destination

7 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet


Host Host

B
Host

Host

packet
switch

F
E

physical link

D
Host

Host Host

Host

Host

Host
Host
Host Host

All packet switches are store-and-forward meaning that entire packets are received,
stored, processed, and then forwarded

The delay experienced by a packet traversing a packet switch, called nodal delay,
consists of four components, namely, transmission delay, propagation delay,
processing delay, and queueing delay:

8 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

10x10
Packet
Switch

- Transmission Delay. If the length of a packet is L bits and the physical link is R bps
L
(bit per second), then it takes Transmission Delay = --- seconds to receive the
R
complete packet
- Propagation Delay. It takes finite amount of time for signal to propagate from point
A to point B. If the distance between A and B is d meters (m) and the signal
propagation speed is v m/s (v is a fraction of speed of light in medium other than the
d
air), then Propagation Delay = --v

9 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet


- Processing Delay. The time required to if detect any errors occurred on the packet,
examine the packets header and determine where to direct the packet. Processing
delays in high-speed routers are typically on the order of microseconds or less.
- Queueing Delay. When there are two or more packets, incoming from different
input ports at the same time, that are destined for the same output port, only one is
allowed to be transmitted and all the others must be buffered waiting in the queue.
The time spent in the queue experience by a packet is called the queueing delay
The amount of queue delay depends on the traffic pattern. Congestion occurs when
the total incoming bit rate of the packets destined for the same output port exceeds
the link rate of the output port. If congestion persists for a prolonged period of time,
the buffer will become full and packets must be dropped
Under non-congestion condition, queueing delay is on the order of microseconds to
milliseconds in practice. Under congestion condition, the delay is infinite if the
buffer size is infinite. In practice, buffer size is finite and queueing delay is
constrained because dropped packets do not account for the delay

For example, let L = 1kByte , R = 1Gbps , the length of a coast-to-coast optical fiber
is 8,000 km (signal moves in fiber at 2/3 of light speed in air), 1 GHz processor and it
takes 100,000 clock cycles to process a packet. An output port of the switch is lightly

10 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet


congested and the average queueing delay is 10 ms. What is the total average delay
through the switch?

8000= 8s ,
Transmission Delay = ----------9
10
8000 ( 1000 )
Propagation Delay = ------------------------------------------------- = 18ms , and
3--- ( 3 ) ( 100000000 )
2
100000
Processing Delay = ------------------ = 100s The average delay through the switch is
1000
Delay = 0.008 + 18 + 0.1 + 10 28ms . If there are ten switches on the path from the
source to destination, then the total average end-to-end delay is 280ms which is really
small unless one or more of the switches are under congestion
We can see that the small end-to-end delays make real-time applications possible in the
Internet

It prioritizes the storage and processing of packets based on the priority encoded in the
packets. This way QoS (Quality of Service) is achieved

It selectively drops packets if congestion occurs and there is no space available to store
packets
11 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

It may do deep packet inspection for marking or dropping unwanted or illegitimate


packets (packet filtering)

The performance of the switches is a critical factor determining the performance of the
network

The number of inputs and outputs ranges from under ten for small switches to
hundreds for large switches

The line speed ranges from 100 Mbps to 100 Gbps (100 billion bits per second)

The total capacity ranges from 1 Gbps to over tens of Tbps (1 tera bits per second or
1012 bits per second)

They cost from under $100 for small layer 2 bridges to over $1M for layer 3 and above
core routers

What is a physical link?

A physical link connects two or more switches together so that the switches can
exchange packets with each other

A physical link can be a copper cable for Ethernet, optical fiber for SONET
(Synchronous Optical NETwork), FTTH (Fiber To The Home), or wireless for 802.11
wifi or LTE (Long Term Evolution) cellular
12 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

SONET such as OC-192 (10 Gbps) is deployed in high speed long haul transport in the
core network (OC stands for Optical Carrier)

What is a host?

A host can be anything that connects to the Internet such as computer, smartphone,
server, web enabled appliances such as set top box, smart TV, and game console, etc.

Any networked devices in Internet of Things (IoT)

User applications are running in the hosts

To satisfy applications such as web surfing, email, video streaming and conferencing,
online gaming, online shopping, etc., hosts communicate with each other by
exchanging packet via the Internet in which one or more hosts generate packets and
one or more hosts consume the packets

There are hundreds of millions of hosts connected to the Internet and billions of users

You may consider the hosts as part of the Internet or not

What is a protocol?

A protocol is set of procedures that defines (1) the format, (2) meaning, and (3) the
actions taken by the hosts or switches

13 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

Protocol are used (a) as a common language for exchanges messages, (b) for reliable
data transfer, (c) for controlling congestion, (d) for exchanging routing information, (e)
exchanging user messages, etc.

What is an Intranet?

A private Internet, including all its attached resources, that can only be accessed by
authorized users. Attached resources include web and content servers, data storage,
printers, computing, document. etc.

It should be noted that the web is not a network but rather just one of many
distributed applications that use the communication services provided by the Internet

14 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet


Protocol Layering

Applications, hardware, software, and control of networks are very complex and we
use a layered approach to facilitate the ease of design, implementation, upgrade, and
maintenance of networks
Application (Layer 7)
Presentation (Layer 6)
Session (Layer 5)

Application (Layer 5)

Transport (Layer 4)

Transport (Layer 4)

Network (Layer 3)

Network (Layer 3)

Data Link (Layer 2)

Data Link (Layer 2)

Physical (Layer 1)

Physical (Layer 1)

OSI 7-Layer Model

Internet 5-Layer Model


15 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

ISO (International Organization for Standardization)s OSI (Open System


Interconnect) uses 7-layer model and Internet uses a 5-layer model

There is a clear distinction between functions executed in different layers but functions
within a layer are similar

The design, implementation, and maintenance of the layers are de-coupled so that a
layer can be modified, deleted, or added without affecting other layers

A layer uses the service provided by the immediate lower layer and provides service to
the immediate upper layer

The interface between two adjacent layers defines what information and services a
layer must provide to its upper layer; the method of information exchange and services
across an interface is implementation dependent

End systems must support all five layers

Intermediate network nodes, the switches, do not support layer 4 and 5

Layer 1, 2, and 3 are the network support layers that actually move data across the
network between peers
If layer 1, peers are the two nodes at the end points of a physical link
If layer 2 or 3, peerr are the end-to-end hosts

16 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

A node supporting layer n must also support layer n-1, n-2, and 1

The network model of the 5-layer model is shown below


end system A

Application
5 - 4 interface

Transport

4 - 3 interface

intermediate node

intermediate node

peer-to-peer application layer protocol


peer-to-peer transport layer protocol

Network

Network

Network

3 - 2 interface

3 - 2 interface

3 - 2 interface

end system B

Application
5 - 4 interface

Transport

4 - 3 interface

Network

3 - 2 interface

Data Link

Data Link

Data Link

Data Link

2 - 1 interface

2 - 1 interface

2 - 1 interface

2 - 1 interface

Physical

Physical

Physical

Physical

physical communication
The Network Model of the 5-Layer Internet Model

Intermediate network nodes are collectively called switches

An intermediate node may support layer 1 only and is known as repeater


17 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet


- Examples are Ethernet repeater, SONET add/drop multiplexer, T1 repeater, etc.

An intermediate node may support layer 1 and 2 only and is known as a bridge
- A bridge knows how to forward data to the next node using the layer 2 destination
address
- Two or more bridges form a bridged network

An intermediate node may support layer 1, 2, and 3 and is known as a router


- A router also knows how to forward data to the next node using layer 3 destination
address
- Two or more routers form a routed network

Data may traverse one or more of the intermediate nodes in order to arrive at the
destination node

Layer 4 support end-to-end transfer of user application data

Layer 5 is the user applications such as the HTTP (HyperText Transfer Protocol),
SMTP (Simple Mail Transfer Protocol), etc.

18 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet


Application Layer

The application layer protocols reside on hosts (user computers) where user network
applications are run. Network applications means any applications that require data
transfer over the Internet, or network in general.

Common Internet application include World Wide Web running the HTTP (Hypertext
Transfer Protocol) protocol, email running the SMTP (Simple Mail Transfer Protocol)
protocol, and reliable file transfer using the FTP (File Transfer Protocol), video
conferencing running the RTP (Real-time Transport Protocol), voice and video
streaming, cloud storage and computing, name service running the DNS (Domain
Name System), etc.

We will refer the application layer unit of data transfer as a message

Application layer protocols use the service provided by the immediate lower layer, the
transport Layer, for exchanging messages with other hosts attached to the Internet

19 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet


Transport Layer

The transport layer provides application layer message transport service between
application endpoints

Two protocols are supported in the transport layer, namely, TCP (Transmission
Control Protocol) for reliable large file transfer and UDP (User Datagram Protocol)
for not-so-reliable small data transfer

TCP provides a connection-oriented service to its applications where the two endpoints
of the application maintain a connection for the entire life of the application

TCP guarantees delivery of application-layer messages to the destination

TCP supports flow control (that is, sender/receiver speed matching)

TCP also breaks long messages into shorter segments

TCP provides a congestion-control mechanism so that a source throttles its


transmission rate when the network is congested

We refer the transport layer TCP unit of data transfer as a segment

UDP protocol provides a connectionless service to its applications

The UDP unit of data transfer is referred as user datagram

20 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

UDP does not provide segmentation service requiring the application layer to pass
messages of length no larger than the maximum user datagram length

UDP is a no-frills service that provides no segmentation, no reliability, no flow control,


and no congestion control services

Transport layer protocols use the service provided by the immediate lower layer, the
network layer, for exchanging segments with other hosts attached to the Internet

Network Layer

We will refer the network layer unit of data transfer as a packet

The Internets network layer is responsible for moving network layer packets known as
datagrams from one host to another

The transport layer protocol (TCP or UDP) in a source host passes a transport layer
segment or user datagram and a destination address to the network layer. The network
layer provides the service of delivering the segment or user datagram to the transport
layer in the destination host

The network layer includes the IP Protocol, which defines the fields in the datagram as
well as how the end systems and routers act on these fields

All Internet components that have a network layer must run the IP protocol
21 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

A host connected to the Internet must has an IP addresses and the IP address uniquely
identifies the hosts in the Internet

A hosts geographical location determines its IP address - group of similar IP addresses


are located in the same area

The purpose of assigning IP addresses to hosts based on the hosts geographical


locations is to facilitate efficient routing of packets allowing the Internet to scale to
todays huge size

Though IP addresses can be statically assigned to hosts, it is much more common to


dynamically assigning IP addresses using the DHCP (Dynamic Host Configuration
Protocol) protocol

The IP address IPv4 (IP version 4) is 32 bits long and IPv6 is 128 bits long. It is
expected that 128-bit address can never run of space

Network layer contains routing protocols that determine the routes that datagrams take
between sources and destinations

Network layer has many routing protocols (called layer 3 routing) such as RIP
(Routing Information Protocol), OSPF (Open Shortest Path First), BGP (Border
Gateway Protocol), etc.

The network administrator can run any routing protocol desired


22 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet


Data Link Layer

Data link layer has two sublayers, namely, LLC (Logical Link Control) and MAC
(Medium Access Control). LLC is closer to the network layer and MAC is closer to the
physical layer

The data link MAC sublayer has its own address space known as MAC address which
is 48 bits long

Every physical network interface has a universally unique MAC address which has no
geographic significant but rather assigned by the manufacturers of the network
interface hardware

We will refer the data link layer unit of data transfer as a frame

Independent of the network layer routing, data link layer can move frames from the
source host to the destination host. The protocol is known as layer 2 bridging

Because MAC address has no geographic significant, layer 2 network is not very
scalable

The services provided by the LLC sublayer depend on the specific link layer protocol
that is employed over the link (the link between the two endpoints of the link, not the
link between two adjacent hosts)

23 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

Some link-layer protocols provide reliable delivery and flow control, from transmitting
node, over one link, to receiving node
- This reliable delivery service is per link, not end-to-end (TCP delivers end-to-end
reliable service)
- Flow control can temporary stop the source from transmitting more frames. For
example, a station on the Ethernet can request all other stations to stop transmission

Examples of data link layer protocols include Ethernet, WiFi, and the cable access
networks DOCSIS protocol

The moving of packets in the Internet relies on both the layer 3 routing and layer 2
bridging. A packet may be switched by different routing protocols and switched by
different bridging protocol. You will learn more about these important switching
technique

The MAC sublayer is also responsible for recovering the frame from the raw bit stream
presented to it from the physical layer

Frames on the raw physical layer bit stream to recover the packet

24 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet


Physical Layer

The physical layer unit of data transfer is a bit

The physical layer coordinates the functions required to move individual bits within a
layer 2 frame from one node to the next over a physical medium

The protocols in this layer are link dependent and further depend on the actual
transmission medium of the link (for example, twisted-pair copper wire, single-mode
fiber optics)

For example, Ethernet has many physical-layer protocols: one for twisted-pair copper
wire, another for coaxial cable, another for fiber, and so on. In each case, a bit is moved
across the link in a different way

It deals with mechanical specification


- Examples are Ethernet RJ45 connector, SC/ST type optical connectors

It deals with electrical


- Examples are voltage level, pulse shape, surge protection, etc.

It deals with synchronization


- Examples are clock recovery, jitter/wander, etc.

25 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

It deals with transmission medium, bit rate and accuracy, etc.

Encapsulation

A hosts application layer generates messages and passes the message down to the
transport layer for delivery

The transport layer divides the message into multiple fragments

The transport layer encapsulates the fragments, one at a time, with a transport layer
header and trailer to form layer 4 segments, which is then passed down to the network
layer for delivery

The network layer encapsulates the segments with a network layer header and trailer to
form layer 3 packets, which is then passed down to the network layer for delivery

The data link layer encapsulates the packets with a data link layer header and trailer to
form layer 2 frames, which is then passed down to the physical layer for delivery

At one end of physical link, the physical layer transmits the layer 2 frames bit-by-bit to
the other end-point of the link

The figure below shows the entire process of encapsulation of the five layers

Decapsulation
26 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet


Application
Transport

Message

TL header

Network

NL header

TL header

Data Link

DL header

NL header

Physical

Message

Message

TL header

TL CRC

TL CRC

NL CRC

Message

TL CRC

NL CRC

DL CRC

send DL frame bit by bit over the communication link

The node, at the other end of physical link, the physical layer receives and recovers the
bits and passes them to the data link layer

Case I: The node is a layer 2 bridge

(a) Frames on the raw bit stream to recover the MAC frame

(b) Extracts the destination MAC address from the frame

27 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

Uses the extracted MAC address to look up the layer 2 forwarding table for the output
link to forward the frame (we will study how the layer 2 forwarding table is created
later in this course)

Case II: The node is a layer 3 router

(a) and (b)

(c) If the extracted MAC address matches the MAC address of the nodes network
interface, from which this frame was received, then
Strips off the layer 2 header and trailer (decapsulate) of the MAC frame to obtain the
IP packet and passes it to the network layer, else
drops the MAC frame (because the frame is not intended for this node)

Extracts the destination IP address from the packet and users it to look up the routing
table for the output link to forward the packet (we will study how the routing table is
created later in this course)

Case III: The node is a host

(a), (b), and (c)

The hosts IP address should match the destination IP address contained in the packet

28 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

Strips off the layer 3 header and trailer of the packet to obtain the layer 4 segment and
passes the segment to the transport layer

If this segment is the last segment of the message,


Reassembles the message and passes it to application layer, else
Saves the segment together with the other saved segments belonging to the same
message

The figure below shows the action of encapsulation and decapsulation

Application

encapsulation
decapsulation

Transport

Application
Transport

Network

Network

Network

Network

Data Link

Data Link

Data Link

Data Link

Physical

Physical

Physical

Physical

29 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet


Connection Models
Path Connection-Oriented (CO) Service

The physical path between the source S and destination D is pre-established before
they exchange data
switch

B
q
host S

A
r

link

t
s D

C
u

y F z
x
w E

host D

In the figure, the path is


Host S sends packets to switch A via As input port q
Switch A directs the packets, out of its output port r, to switch D via Ds input port s
30 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet


Switch D directs the packets, out of its output port t, to switch C via Cs input port u
Switch C directs the packets, out of its output port v, to switch E via E input port w
Switch E directs the packets, out of its output port x, to switch F via Fs input port y
Switch F directs the packets, out of its output port z, to host D

Packets must identify to the switches which path they belong to and this is
accomplished by the path identifier field in every packet

Packets having the same path identifier traverse hop-by-hop across the switches along
the same established path to the destination

Each switch on path maintains the path information such as the input to output port
mapping, the bandwidth and delay requirement, etc.

When host S and host D cease the conversation, the path must be teared down either
explicitly or by time-out

If tear down explicitly, S (D) sends tear-down messages down the path so that the
switches on the path remove the state information about the path

If tear down by time-out, a switch on the path removes the state information about the
path if it has not seen any data on the path for a predetermined amount of time

31 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

The major use of path CO service is to provide QoS (Quality-Of-Service) such as


reserved bandwidth, delay guarantee, packet dropping guarantee, and buffer space
guarantee
This type of guarantees was originated from the telcos like AT&T, Pacific Bells, etc.
because their circuit switched POTS (Plain Old Telephone Service) networks naturally
support the guarantees

Path Connectionless (CL) Service

Switches in the network do not maintain any state information about any path between
two hosts, and therefore, there is no physical path pre-established between the source
and destination hosts

The destination address must be included in every packet sent by the source

Switches in the network forward packets hop-by-hop based on the destination address
of the packets

Packets originated from the same source may take different routes to the destination

Switches may prioritize packets based on the priority encoded in the packets to provide
some QoS, but not as strong as the path CO service

The Internet is basically a path connectionless network

32 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

Path CO and path CL are switching services provided by the network

End-to-end CO and end-to-end CL are connection services provided by the hosts

End-to-End Connection-Oriented Service

Prior to data transfer, two hosts use control packets to exchange control information to
establish a connection first. The transport layer TCP (Transmission Control Protocol)
is an excellent example of this service in the Internet

The source and destination hosts maintain the state information (the control
information) about the connection

End-to-end CO service guarantees packets transmitted from a sender will eventually be


delivered to the receiver without errors, in the order they are sent, and in their entirety

Delivery without errors is by means of error detection and retransmission

Delivery in the order the packets are sent is by means sequence numbers, buffering,
and retransmission

Delivery in entirety is by means of segmentation, reassembly, and sequence number

While the major use of the end-to-end CO service is to provide very reliable data
transfer, the major use of path CO service is to provide QoS such as low delay, delay
variation, packet loss
33 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

End-to-end CO service may be supported by path CO or path CL service

An example of end-to-end CO connection supported by path CO service is ATM which


is mainly used by the telephone companies

In end-to-end CO service supported by path CL service, the switches in the network do


not aware of the opened end-to-end user connections between hosts, and therefore, no
state information about the user connection is maintained in the switches

An example of end-to-end CO connection supported by path CL service is the well


known TCP protocol used in the Internet

End-to-end Connectionless Service

The source host wishes to send packets to a destination host, simply sends the packets
containing the address of the destination host without pre-establishing an end-to-end
connection

If packets are dropped in transit to the destination host, the hosts would not know or
more precisely the transport layer used in the Internet would not know

If packets arrive to the destination host out-of-order, the transport layer also would not
know

34 of 35

EECE 7902 / 8902 Advanced Computer and Wireless Networks

8/23/2016

Computer Networks and the Internet

Capabilities can be built right into the applications to detect packet loss and out-oforder arrival defects

End-to-end CL service may be supported by path CO or path CL service

An example of end-to-end CL connection supported by path CO service is video


teleconferencing since it requires QoS but can tolerate occasional delay or frame loss

An example of end-to-end CL connection supported by path CL service is the well


known UDP protocol for the user datagram service used in the Internet

Some video streaming applications use user UDP across the Internet because it neither
requires QoS not high reliability

35 of 35

You might also like