Transcript Document

Chapter 5
Switching
Introduction
• Look at:
– Bridges and Bridging(5.1)
– Switches and Switching(5.2)
– Spanning Tree Algorithm(5.3)
– Virtual Local Area Networks (VLANs)(5.4)
– Integrating Switches(5.5)
Introduction
• Moving electronic signals from one
interface to another is the term used to
describe switching
• Ethernet networks started as a single
segment bus topology with all clients
able to receive the electronic signals of
all other clients on their network
Bridges and Bridging
• A bridge operates by simply plugging it into
the wall and connecting network devices to it
• Bridging can isolate groups of network
devices by segmenting the wire signals of
one group from another group
• Each of the separate groups is referred to as
a collision domain
• Fewer devices competing per network
segment mean better communication
Bridges and Bridging
• There are several different
implementations of bridging:
– Transparent bridging
– Source route bridging
– Translational bridging
• Each has a specific purpose in
controlling traffic flows and collisions
Bridges and Bridging
• In transparent bridging, the network devices
are unaware of the presence of the bridge
• The bridge forwards traffic based on a first-in,
first-out (FIFO) basis and combines the
transmissions allowing multiple clients to
communicate with the Server
• Network transmissions take a small amount
of time, and the bridge must be fast enough
to keep up
Bridges and Bridging
• Source route bridging (SRB) is designed for
use with Token Ring networks
• In Token Ring networks, rings and bridges
have an assigned numeric value
• The client determines the best path through
the Layer 2 network
• The default behavior is to use the path
identified in the first response received
Bridges and Bridging
• Translational bridging contains a mixture of
Ethernet and Token Ring clients
• Very few standards exist for translational
bridging, and implementations vary between
vendors
• Some translational bridging implementations
may not work between vendors
• It’s important to research compatibility if you
plan on implementing translational bridging
Switches and Switching
• Switching offers the following
improvements over bridging:
– Higher port density
– Faster packet processing capabilities
– Quality of service (QoS) capabilities
– Use of virtual LAN (VLAN) technology
• Each port is a single bridging device
creating its own collision domain
Switches and Switching
• Switches achieve faster packet
processing capabilities than bridges in a
number of different ways:
– Application-specific integrated circuits
– Better CPUs
– Cut-through switching
– Fragment-free switching
Switches and Switching
• Circuit switching is the oldest form of
switching
• It establishes calls over the most efficient
route available at the time
• It can be wasteful compared to other types of
communication
• It is wasteful because the circuit remains
active even if the end stations are not
currently transmitting
Switches and Switching
• Examples of circuit-switched networks
include:
– Asynchronous Transfer Mode (ATM)
– Integrated Services Digital Network (ISDN)
– Leased digital line
– T-1
– Analog dial-up line
Switches and Switching
• Types of switching include:
– Time-division which is the switching of
time-division multiplexed (TDM) channels
by shifting bits between time slots in a TDM
frame
– Space-division in which single
transmission-path routing is accomplished
using a switch to physically separate a set
of matrix contacts or cross-points
Switches and Switching
• Types of switching include:
– Time-space division which precedes each
input trunk in a crossbar with a TSI, and
delays samples so that they arrive at the
right time for the space-division switch’s
schedule
– Time-space-time division is where data
passes through a space switch to create
circuits for TDM outlets
Switches and Switching
• Packet switching is implemented by protocols
that rearrange messages into packets before
sending them
• Each packet is then transmitted individually
• Because packets are send via different
routes, they may not arrive in the order in
which they were originally sent
• Technologies such as Cellular Digital Packet
Data and Voice over IP is currently
implementing voice communication using
packet switching
Switches and Switching
• In packet switching protocols, a port is
represented by a value between 1 and 65535
• The port number indicates the type of packet
• Port mapping, or port address translation is a
process where packets arriving for a
particular socket can be translated and
redirected to a different socket
• Port mapping is necessary only for incoming
transmissions, not for returning traffic
Switches and Switching
• Blocking occurs when a network is unable to
connect stations to complete a circuit
• In packet switching, blocking is caused by
congestion
• Congestion management is used to help
ensure that the blocked packet is not lost
• Congestion management is implemented
using buffers or queues
Switches and Switching
• Asynchronous transfer mode (ATM) is a highspeed, connection orientated packet
switching technique
• It uses short, fixed-length packets called cells
• ATM is connection oriented
• The switching elements have pre-defined
routing tables to minimize the complexity of
single switch routing
Switches and Switching
• Switching fabric is the combination of
hardware and software that transfers data
coming into a node to the appropriate port on
the next node
• The bandwidth of a switching fabric is defined
by its data width in bits and the speed at
which it can transmit these bits
• Switching fabric includes data buffers and
shared memory
Switches and Switching
• A crossbar switch is a device that directly
switches data between an input port and an
output port without sharing a bus with any
other data
• Crossbar topology is similar to bus topology
• There is only one path that all devices share
• A crossbar switch environment offers more
flexibility and greater scalability
Switches and Switching
• A broadcast connection is when a device
simultaneously sends data to more than one
receiving device
• A broadcast connection in a non-blocking
multistage network that has an input port
connected simultaneously to several unused
output ports
• The network must be non-blocking for
broadcast assignments
Switches and Switching
• A switching element is the basic building
block of a switch
• Switching fabric consists of switching
elements that facilitate a particular switching
mechanism
• When a large number of switching elements
are connected together in a network, this is a
multistage interconnection network (MIN)
Spanning Tree Algorithm
• The Spanning Tree Algorithm (STA) was
created to overcome the weakness inherent
in bridges
• This weakness is the flooding behavior of
broadcast, multicast and unknown traffic
types
• Problems occur with broadcasts when you
configure two or more bridges in a loop
Spanning Tree Algorithm
• By interconnecting multiple switches with
redundant paths, you overcome problems
with faulty cables or port failures
• Another less well-known side effect of a loop
is the corruption of the forwarding tables on
all the switches
• It would take very little time before Layer 2
broadcast loops completely destroy the
functionality of a network
Spanning Tree Algorithm
• The Spanning Tree Protocol (STP) prevents
bridging loops by identifying a preferred path
through a series of looped bridges
• Administrators can provide redundancy and
fault-tolerance by wiring a loop, and then
using STP turn off ports that would cause
loops to occur
• If a primary link fails, STP will reactivate the
back-up port allowing normal operation of the
network to continue
Spanning Tree Algorithm
• The STP process is accomplished by:
– Election of a root bridge
– Each bridge, when first turned on, begins
the election process by sending a packet
called a bridge protocol data unit (BPDU)
– Each bridge believes itself to be the root
until it receives a BPDU with a lower root
bridge ID
Spanning Tree Algorithm
• The STP process is accomplished by:
– Identification of a root port
– Once the root bridge is determined, each
non-root bridge finds the best path to the
root using path cost
– Path cost is a numeric value that bridges
use to determine the preference of a given
path. It is derived by taking 1000 MBps and
dividing it by the speed of the link
Spanning Tree Algorithm
• The STP process is accomplished by:
– Identification of a designated port
– After the switches determine the root
bridge and the best path to the root, the
designated port is determined
– When this step finishes, the loop is broken
because one of the switches is blocking
traffic on one of it’s ports
Spanning Tree Algorithm
• The switch continues to receive traffic
on the port and evaluates each frame
• All user traffic is filtered including
unicasts, broadcasts, and multicasts
frames
• Anytime a new switch is plugged in, the
process of determining the root bridge
occurs
Spanning Tree Algorithm
• Most bridges and switches go through the
following phases when activating ports to
determine which system is the root bridge
and which ports are active or disabled
– Blocking
– Listening
– Learning
– Forwarding
Virtual Local Area Networks
• Prior to the concept of VLANs the only way
for administrators to block broadcast traffic
was to implement a router
• VLANs represent a solution allowing
administrators to group ports on their
switches so that broadcast traffic is only
passed among ports within the group
• Once a VLAN is established the ports will
only flood broadcast traffic to their own
members
Virtual Local Area Networks
• Once the VLAN is created, it is
necessary to have a Layer 3 router
• The clients on separate VLANs will no
longer communicate with each other
• The network devices can no longer
discover each other by using ARP
broadcasts
Virtual Local Area Networks
• The basic process of creating a VLAN
involves tagging the inbound packet
with a VLAN ID
• Tagging is accomplished in one of two
ways:
– Frame insertion
– Frame encapsulation
Virtual Local Area Networks
• Frame insertion involves inserting a small
identifier into the frame as it is received at the
switch’s port
• In Frame encapsulation, the entire frame is
simply encapsulated inside a VLAN ID
header and checksum
• Encapsulation is still a proprietary method of
VLAN tagging but is gaining in popularity
• It is fast at frame insertion and less prone to
damaging the underlying frame
Virtual Local Area Networks
• VLAN trunking uses the VLAN concept
and applies it to more than one switch
• The solution for routing frames is trunking or
making one port forward traffic for all VLANs
• Doing so cuts port usage down considerably
• While VLANs appear complex at first, they
offer a great deal of flexibility for
administrators
Virtual Local Area Networks
• If a particular device needs to be moved
from one broadcast domain to another,
only the port configuration on the switch
needs to be changed, not the physical
location of the device
• Fewer ports are used on the routers,
saving money and configuration time
Integrating Switches
• Integrating hubs and switches provide a
migration path as networks are upgraded
from hub environments to switching
topologies
• This can be done by connecting a crossover
network cable from one of the hub ports into
a switch port
• If the hub or switch includes a crossover port
then you use a regular networking cable
Integrating Switches
• All ports on a hub are in the same broadcast
and collision domain
• All ports on a switch are in the same
broadcast domain but each port is its own
collision domain unless VLANs are used
• Connect your servers and other high traffic
workstations directly to a switch so that they
are free from collisions
Integrating Switches
• Some switches can allow you to add
functions such as:
– Packet filtering
– Encryption
– Auditing/Accounting
– Tunneling
– Routing
Integrating Switches
• By adding these directly to the switch, frames
are copied into the memory of the switch
once and then forwarded on to the next
destination without further copying
• This allows a switch to route frames at wire
speed
• An additional side benefit of integrating
switches with routing technology is the
simplification of network design