Chapter 8

Switching

Copyright The McGraw-Hill Companies, Inc. Permission required for reproduction or display.

How do we connect devices?

1. Connect each pair of end nodes using point-to-point connections
2. Creating temporary connections between end nodes through a
network of switches
Benefits of Switching:
Fewer physical links
More scalable
On-demand connection
Better network utilization

8.2

Taxonomy of switched networks

8.3

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.
However, each connection uses only one dedicated
channel on each link. Each link is normally divided
into n channels by using FDM or TDM.
Topics discussed in this section:
Three Phases
Efficiency
Delay
Circuit-Switched Technology in Telephone Networks
8.4

A trivial circuit-switched network

In circuit switching, the resources need to be reserved during
the setup phase.
The resources remain dedicated for the entire duration of data
transfer until the teardown phase.

8.5

Example 8.1: Circuit-Switched Operation

Here Telephone 1 is connected to telephone 7; 2 to 5; 3 to 8; and 4
to 6.
Multiplexing techniques are used to carry several circuits on a
trunk

8.6

Example 8.2: Circuit Switching for Computers

Consider a circuit-switched network that connects computers in
two remote offices of a private company. The offices are
connected using a T-1 line leased from a communication service
provider. There are two 4 8 (4 inputs and 8 outputs) switches in
this network. For each switch, four output ports are folded into
the input ports to allow communication between computers in the
same office. The other four output port connect the branches.

8.7

Data Transfer in a Circuit-Switched Network

Before Starting communication, a set-up phase is required to
reserve the resources for the circuit.
Data transfer is a continuous flow (not packetized)
Once the transfer is done, a tear-down phase is required to
release resources.

8.8

Resource Reservation
Examples of resources that must be reserved during the set up
and released during the teardown:
Channels
Frequency band for FDM
Time Slots in TDM
Switch buffer space
Switch CPU time
Switch input-output ports

8.9

Efficiency and Delay

Main benefit of Circuit Switching: Guaranteed Quality of
Service
No competition over resources; if not available at set up time,
the call will be denied.
Efficiency: Not efficient because resources are wasted if not
used (no sharing)
Delay: Additional delay at the beginning and end of
communication for set up and teardown, but delay during data
transfer is minimal
No routing table look up
No queuing or waiting time
8.10

Applications of Circuit Switching

Public Switched Telephone Network (PSTN)
Pre-3G Cellular Networks (e.g. GSM)
Optical Mesh Networks and SONET
X.21 (Physical layer for X.25 networks)
ISDN

8.11

8-2 DATAGRAM NETWORKS

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
network and the governing protocol.
Topics discussed in this section:
Routing Table
Efficiency
Delay
Datagram Networks in the Internet
8.12

A datagram network with four switches (routers)

In a packet-switched network, there is no resource reservation;
resources are allocated on demand.
Typically more efficient than circuit switching in resource
handling (though overhead must be added to each packet)
Quality of Service cannot be guaranteed

8.13

Routing table in a datagram network

A switch in a datagram network
uses a routing table that is based
on the destination address
The destination address in the
header of a packet in a datagram
network remains the same during
the entire journey of the packet

8.14

Delay in a datagram network

No set-up delay, but packets experience processing and queuing
delays during the data transfer.
In the following if the transmission time of the packet is T and
the propagation time on each hop is and waiting time at each
node is w, the total end to end delay for this packet is 3T+3+2w

8.15

Applications of datagram network

The Internet
Any packet switching network in general
3G and 4G Cellular networks
VoIP

8.16

8-3 VIRTUAL-CIRCUIT NETWORKS

A virtual-circuit network is a cross between a circuitswitched network and a datagram network. It has
some characteristics of both.
Topics discussed in this section:
Addressing
Three Phases
Efficiency
Delay
Circuit-Switched Technology in WANs
8.17

Virtual-circuit network
In Virtual Circuit Networks the Data is Packetized
An End-to-end path is established (setup and teardown) so that
all packets follow the same route.
Often implemented in data link layer

8.18

Virtual-circuit identifier
Two levels of addressing exist in Virtual Circuit networks:
1) Global Addressing for End hosts (e.g. IP Addresses)
2) A local Virtual Circuit Identifier (VCI) on a switch that
identifies the incoming and outgoing circuits

8.19

Switch and tables in a virtual-circuit network

The forwarding table maps the incoming and outgoing VCIs,
instead of destination address and next hop which were used in
datagram networks.

8.20

Source-to-destination data transfer in a VC network

8.21

Setup request in a virtual-circuit network

a) Node A sends a setup request to switch 1 for a connection to node B.
b) Switch 1 finds the next hop to reach host B either through manual
configuration or routing table look up (only during setup). Switch 1
assigns an incoming VCI, and forward to switch 2
c, d) Switch 2 and 3 do the same until the set up request arrives at B.
e) Node B accepts
the request and
assigns a VCI.

8.22

Figure 8.15 Setup acknowledgment in a virtual-circuit network

All the nodes then communicate this information back through a
setup acknowledgment message and update their forwarding tables.

8.23

Note

In virtual-circuit switching, all packets

belonging to the same source and
destination travel the same path;
but the packets may arrive at the
destination with different delays
if resource allocation is on demand.

8.24

Delay in a virtual-circuit network

8.25

Applications of virtual-circuit network

Used in switched WANs
Frame Relay
ATM
MPLS

8.26

8-4 STRUCTURE OF A SWITCH

We use switches in circuit-switched and packetswitched networks. In this section, we discuss the
structures of the switches used in each type of
network.

Topics discussed in this section:

Structure of Circuit Switches
Structure of Packet Switches

8.27

Crossbar Switch
The Crossbar switch is a Space-Division Switch.
N input, N out put and N2 crosspoints
Enable crosspoint (i , j) when connecting input i to output j
Only one path exists for each connection, so it is non-blocking
(two paths never cross at the same point)
Not Scalable

8.28

Multistage switch
Combines several crossbar switches
Requires fewer crosspoints but probability of blocking exists
2
N

For a three-stage switch, the total number of crosspoints is 2kN k

n
which is less than N2

8.29

Example 8.3
Design a three-stage, 200 200 switch (N = 200) with
k = 4 and n = 20.
Solution
In the first stage we have N/n or 10 crossbars, each of size
20 4. In the second stage, we have 4 crossbars, each of
size 10 10. In the third stage, we have 10 crossbars,
each of size 4 20. The total number of crosspoints is
2kN + k(N/n)2, or 2000 crosspoints. This is 5 percent of
the number of crosspoints in a single-stage switch (200
200 = 40,000).
8.30

Clos Criterion
In order to avoid blocking in a multistage switch, the
number of middle stage switches must be at least 2n-1.
Clos Criterion:

k 2n 1

The following value for n minimizes the number of

crosspoints for the multistage switch (from slide 8.29):
1
2

N
n
2
8.31

N
2

# of Crosspoints 4 N 2 N 2 1

Example 8.4
Redesign the previous three-stage, 200 200 switch, using the Clos
criteria with a minimum number of crosspoints so that blocking
does not occur.
Solution
We let n = (200/2)1/2, or n = 10. We calculate k = 2n 1 = 19. In the
first stage, we have 200/10, or 20, crossbars, each with 10 19
crosspoints. In the second stage, we have 19 crossbars, each with
(N/n) = 20 inputs and 20 20 crosspoints. In the third stage, we
have 20 crossbars each with 19 10 crosspoints. The total number
of crosspoints is 20(10 19) + 19(20 20) + 20(19 10) = 15200.

8.32

Time-Slot Interchange Switch

This is a Time-Division Switch that combines a TDM multiplexer
and a TDM demultiplexer.
The switch controls which input is connected to which output at
each time slot.
Needs no crosspoints but adds delay (waiting until its your turn).

8.33

Time-Space-Time (TST) switch

Combination of Time-Division and Space-Division Switching

8.34

Packet Switches
Packet switches have a different architecture from a circuit switch.
The switching fabric can be a circuit switch (e.g. crossbar) or other
switch fabrics that well discuss in the following.

8.35

Input and Output Ports

8.36

Banyan MultiSwitch Fabric

For N inputs, we have log2 N stages and n/2 microswitches at each
stage.
Routing is done based on the bit string of the output port number
(see the next slide)

8.37

Examples of routing in a banyan switch

Problem with banyan switch: possibility of internal collision when

two paths going to different ports cross paths at an intermediary
microswitch.
8.38

Batcher-banyan switch
Solves the internal collision problem by adding a trap module.
The trap module checks the output ports of the packets and
determines if there will be an internal collision. If so, it will schedule
packets for different time slots.

8.39