214 CHAPTER 8 SWITCHING

Traditionally, three methods of switching have ben important: circuit switching.

packet switching, and mcssage switching. The frst two are commonly used today. Thc
third has been phased out in general communications but still has networking applications.
We can then divide today's networks into thrce broad catcgorics: circuit-switched networks,
packet-switched nctwerks, and messagc-switched. Packet-switchcd nctworks can further
bc divided into two aubcatcgorics,-virtual-circuit networks and datagram networks
· as shown in Figure 8.2.

Figure 8.2 Tuxpnomy of swilched networks

Switchcd
ntwurks

Packet-vwitchcd Mcssagc-swichcd
Circuii-switched nctworks
nclworks networks

Vitual-circuit
Daiagrann
networks nei works

some common charactcristics

We can say that thc virtual-circuit networks havc
discusscircuit-switched
with circuit-switched and datagram networks. Thus, we first
networks, then datagrain networks, and finally virtual-circuit networks.
and virual
Today the tendency in packet switching is to combine datagram networks idca,
circuit networks. Networks route thc first packet bascd on the datagram addressing
packets coming from the same
but then create a virtual-circuit nctwork for the rest of the
networks in future
source and going to the same destination. We will see somc of these
chapters. it to the
In message switching, each switch stores the whole message and forwards
next switch. Alhough, we don't see message switching at lower layers, it is sill used in
some applications like electronic mail (e-mail). We will not discuss this topic in this book.

8.1 CIRCUIT-SWITCHED NETWORKS

A circuit-switched network consists of a set of switches connected by physical links.
A
connection between two stations is a dedicated path made of one or more links. How
evcr, cach connection uses only one dedicated channel on cach link. Each link is nor
mally divided into nchannels by using FDM or TDM as discussed in Chapter 6.
Acircuil-switched network is made of a set of switches connected by physical links,
in which each link is divided into n channels.

Figure 8.3 shows a trivial circuit-switched network with four switches and four
links. Each link is divided into n(n is 3in the figure) channels by using FDM or TDM.
 SECTION 8.| CIRCUIT-SWITCHED NETWORKS 215

Figure 8.3 Atrivial ciruit-nvilched nework

One link.n channcls

Path

We have explicitly shown the multiplexing symbols to emphasize the division of the
link into channcis even though multiplexing can be implicitly included in the switch
fabric.
The end systems, such as computersS or telephones, are directly conaccted to a
switch. We have shown onlytwo end systems for sin1plicity. When end system Anceds
to communicate with end system M, system Aneeds to request aconnection to Mthat
must be acccptod by all switches as well as by Mitself. This is caled the setup phase;
a circuit (channel) is reserved on each link, and the combination of circuits or channels
defincs the dedicated path. After the dedicated path made of connected circuits (channels)
is established, data transfer can take place. After all data have been transferred, the
circuits are torn down.
We need to ermphasize several points here:
OCircuit switching takes place at the physical layer.
O Before starting communication, the stations must make a reservation for the resources
o be used during the communication. These resources, such as channels (bandwidth
in FDM and time slots in TDM), switch buffers, switch processing time, and switch
inputoutput ports, must remain dedicated during the entire duration of data transfer
until the teardown phase.
0 Data transferred between the two stations are not packetized (physical layer transfer
of the signal). The data are acontinuous flow sent by the source station and received
by the destination station, although there may be periods of silence.
OThere is no addressing involved during data transfer. The switches route the data
based on their occupied band (FDM) or time slot (TDM). Of course, there is end-to
end addressing used during the setup phase, as we will sce shortly.

In cireuit switching, the resources need to be reserved during the setup phase;
the resources remain dedicated for the entire duration
of data transter until he teardown phase.
216 CHAPTER 8 SWITCHING

Example 8.1
nctwork to conncct cight tclcphoncs in a small
As a trivial cxamplc. Jct us usc a circuit-switchcd
channcls. We assumc that cach link usCs FDM (o
arca. Communication is through 4-kHz voicc
bandwidth of cach link is then 8 kHz. Figurc 8.4
Conncct a maxi1num of two voice channcls. The 6. Of coursc
7; 2 to 5;3 to 8; and 4 to
shows thc situaion. Tclcphonc lis conncclcd to tclcphonc
madc. Thc switch contyols thc conncctions.
arc
thc situation may change whcn ncw conncctions

Figure 8.4 Cirrit-1vitched network used in Exunmple 8./

Cincuit-swiched etwurk

kHz. kH%

3
0 8
kHz
4
KH%. kHz.
KHz. kHz

Example 8.2
nclwork that connccls computcrs in (wo renolc
As another cxamplc. considcr a circuit-swilchcd
a T-I linc leascd irom a communi
officcs of a privatc company. The offccs arc connccted using
outputs) switchcs in this nctwork. For
cation servicc provider. Therc arc two 4 x8 (4 inputs and 8 to allow comnmunication bctwcen
cach switch, four (output pors arc foldcd into hc input ports
communicalion between the two
computcrs in the samc ofice. Four other output ports allow
offices. Figure 8.5 shows the situation.

Figure 8.5 Circui-Switched nehvork used in Exanple &.2

Circujt-swilched nctwork

4x3 ;4x8
Swich T1 lioe with switch
1.$44 Mhps
 SECTION 8.J CIRCUIT-SWITCHED NETWORKS 217

Three Phases
The actual communication in a cicuit-swilchcd nctwork rcquires thrcc
tion sctup, data transfer, and connection tcardown. phases: conncc

Setup Phase
Before the two partics (or nultiple parlies in a conferencc call) cah
conmunicate, a,
dedicated circuit (combination of channcls in links) necds to be established. Thc cnd sys
tems are normally connectcd through dcdicated lines to the switches, so
conneclion sctup
meana' creating dedicated channels betwéen the 'swithës: For examplc, in Figure
when system A nceds to conncct to system M, it scnds a setup rcquest that
8.3;
includes the
address of system M, tÍ switch 1. Switch I finds a channel between itsclf and switch IV
that can be dedicated for this purpose. Switch Ithen sends the
which finds a dedicated channel between itself and switch II. Switch request to switch IV,
llI informs sys
tem M of system A's intention at this time.
In the next step to Nhaking a connection. an
acknowledgment from system M needs
to be sent in the opposite direction to system A. Only after
system A receives this
acknowledgment is the conncction established.
Note that end-to-cnd addressing is required for creating a connecion
two cnd systems. These can be, for example, the addresses of the belwccn the
by the administrator in a TDM network, or telephone computers assigned
numbers in an FDM network.
Data Transfer Phase
After the establishnent of the dedicated circuit (channels), the (wo
parties can transfer data.
Teardown Phase
When onc of thc partics nccds to disconnccl, a signal is sent to each
the resources. switch lo relcasc

Eficiency
It can be argued that
circuit-switched networks are not as cfficient as the other (wo
types of nctworks because resourccs are allocated during the entire
duration of the con
neçtion. These resoUrces arc unavailable to other connections. In a
telephonc
people nornally terminate the communication when they havc finished thcir network,
However, in computcr nctworks, a computcr can be connccted to another conversation.
evcn if therc is no activity fora long ime. n this case, allowing computer
mcans that other conncctions ure deprived. resources to be dedicated

Delay
Although a circuit-swithed nctwork normally has low
of network is minimal. During data iransfer the data areefficiency, the delay in this type
not delayed at cach
rcsources are allocatcd for the duration of the connection. Figure 8.6 shows switch; the
the idea of
delay in a circuit-switched nctwork when only two switches are
"As Figure 8.6 shows. there is no waiting time at involved.
lo the time necded to crcatc the conncction. transfer each switch, The total delay is due
data, and disconncct the circuit. The
218 CHAPTER 8 SWITCHING

Figure 8.6 Delay in a circui-switched newvork

Conncct

delay
Tolal

Dala iransfer

DIS

Tinic Timc
Tine Time

propagation time of the source

delay caused by the setup is the sum of four parts: the transfer time (height of
request signal
computer request (slope of the firstgray box), theacknowiedgment from the destination
the first gray box),the propagation time of the
transfer time of the acknowledg
computr (slope of the second gray box), and the signal
due to data transfer is the sum of two
ment (height-of the second gray box). The delay
box) and data transfer time (height of
parts: the, propagation time (slope of the coloredthird
The box shows the time needed to tcar
the colored box), which can be very long.
receiver requests disconnection,
down the circuit. We have shown the case in which the
which creates the maximum delay.

Circuit-Switched Technology in Telephone Networks

As we will see in Chapter 9, the telephone companies have
previously chosen the circuit
switched approach to switching in the physical layer; today the tendencyused is moving
number is as the
toward other switching techniques. For example, the telephone teardown
global address, and a signaling system (called SS7) is used for the setup and
phases.

Switching at the physical layer inthe traditiönal telephone

network uses the circuit-switching approach.

8.2 DATAGRAM NETVWORKS

In data communications, we need to send messages from one end system to another.If
the message is going to pass through a packet-switched network, it needs to be divided
into packets of fixed or variable size. The size of the packet is determined by the net
work and the governing protocol.
In packet switching, there is no resource allocation for a packet. This means that
there is no reserved bandwidth on the links, and there is no scheduled processing time
 SECTION 8.2 DATAGRAM NETWORKS 219

for each packet. Resources arc allocated on demand. The allocation is done on a first
come, first-served basis. When aswitch receives a packet, no matter what is the source
or destination, the packet must wait if there arc other packets being processed. As with
other systems in our daily life, this lack of reservation may create delay. For exarmple, if
we do not have a rescrvation at a restaurant, we might have to wait.

In a packet-switched network, there is no resource reservafion;

resources are allocated on demand.

In a datagram network, each packet is treated independently of allothers. Even if

apacket is part of a multipscket transmission, the network treats it as though it existed
alone. Packets in this approach are referred to as datagrams.
Datagram switching is normally done at the network layer. We briefly discuss
datagram networks here as a comparison with circuit-switched and virtual-circuit
switched networks. In Part 4 of this text, we go into greater detail.
Figure 8.7 shows howthe datagram approach is used to deliver four packets from
station A tO station X. The switches in a datagram network are traditionally referred to
as routers. That is why we usc a different symbol for the switches in the figure.

Figure 8.7 Adatagram nertwork with four switches (routers)

Dalagran1 actwork
A

In this example, all four packets (or datagrans) belong to the same message, but
may travel different paths to reach their destination. This is so because the links may be
involved in carrying packets from other sources and do nÍt have the necessary bandwidth
available to catry all the packets from Ato X. This approach can cause the datagrams of
a transmission to arrive at their destination out of order with different delays between the
packets. Packets may also be lost or dropped because of a lack of resources. In most
protocols, it is the responsibility of an upper-layer protocol to reorder the datagrams or
ask for lost datagrans before passing them on to the application.
The datagram networks are sormetimes referred to as connectionless networks. The
term connectionless here means that the switch (packet switch) does not keep information
about the connection state. There are no setup r teardown phases. Bach packet is treated
the same by a switch regardless of its source or destination.
220 CHAPTER8 SWITCHING

Routing Table
Ithere are no sctup or teardown phases, how are the packets routed to their
destinations
in adatagram network? In this type of network, cach switch (or packet switch) has a rout
ing table which is based on the destination address. The routing tables are dynamic and
are updated periodically. The destination addresses and the corresponding forwarding
output ports are recorded in the tables. This is different from the table of a circuit
switched network in which each entry is created when the setup phase is completed and
deleted when the teardown phase is over. Figure 8.8 shows the routing table for a
switch.

Figure 8.8 Routing table in a datagram network

-Destination Outpul:
ddres$: port
1232
4150 2

9130 3

Aswtch in adatagram netvork uses arouing table that is based on the destination address.

Destination Address
Every packet in a datagråm network carries a header that contains, among other infor
mation, the destination address of the packet. When the switch receives the packet, this
destifation address is examined; the routing table is consulted to find the corresponding
port through which the packet shouldbe forwarded. This address,unlike the address.
in a virtual-circüit-switched network, remains the same during the entire journey of the
packet.

The destination address in the header of a packet in ådatagYam network'

R í : remains thesame during the entire journey of thepåcket.

Effciency
The efficiency of a datagram network is better than that of a circuit-switched network;
resources are allocated only when there are packets to be transferred. If asource sends
a packet and there is a delay of a few minutes before another packet can be sent, the
resources can be reallocated during these minutes for other packets from other sources.
 SECTION 8.3 VIRTUAL-CIRCUIT NETWORKS 221

Delay
There may be greater delay in a dalagram nctwork than in å virtoal-circuit network.
Although there are no setup and teardöwn phases, each packet may experience await at a
switch before it is forwarded. In addition, since not all packets in a message necessarily
travel through the same switches, the delay is not uniforin for the, packets of a smessage.
Figure 8.9 gives an example of delay in adaiagram network for one single packet.

Figure 8.9 Delay in adatagram helwork

Transission
thnc
Waitlng delay
Toial
lie

Waiting

Tine Tine Tinte TUk

The packet travels through two swilches. There are three transmission times (37).
three propagation delays (slopes 3t of the lines), and two waiting timcs (wy + w). We
ignore thc processing linc in cach switch. The total dclay is

Toal delay = 3T + 3t + " + w

Datagram Networks in the Internet

As we will see in future chapters, the Internet has chosen the datagram approach to
switching at the network layer. It uses the universal addresses defined in the network
layer to roule packets from the source to the destination.

Svitehing in the Internet is done by using the datagram

approach to packet switching at the network layer.

8.3 VIRTUAL-CIRCUIT NETWORKS

A
virtual-circuit network is across between a circuit-switched network and adatagram
network. [t bas some characteristics of both.
1. As in a circuil-switched network, there are setúp nd teardown phases in addition
to the data transfer phase.