09 switching devices

8/6/2019 09 Switching Devices

1/36

1Prof Nehal N ShahECED, SCET

Computer Communication & NetworkingLecture 4

LAN Technology

&

Switching Device


2/36

Prof Nehal N Shah

ECED, SCET


2

Categories of LAN

LAN type Ethernet ARCnet Token-Ring

Topology Bus/Star Bus/Star Ring

Speed 10/100/1000 2.5Mbps 4/16 Mbps

Mbps

Access method CSMA/CD Token Passing Token Passing

Media type Twisted Pair Twisted Pair Twisted Pair

Coaxial cable Coaxial cable Coa

Optical fiber


3/36

Prof Nehal N Shah

ECED, SCET


3

Ethernet

Ethernet is a baseband LAN specification invented byXerox Corporation that operates at 10 Mbps usingcarrier sense multiple access collision detect

(CSMA/CD) to run over coaxial cable. Any station can access the network at any time. Before

sending data, CSMA/CD stations listen for traffic on the

network. A station wanting to send data waits until itdetects no traffic before it transmits.

10BaseT 10Mbps twisted pair

100BaseT

100Mbps twisted pair


4/36

Prof Nehal N Shah

ECED, SCET


4

Local Area Networks & Devices


5/36

Prof Nehal N Shah

ECED, SCET


5

Wide Area Networks & Devices


6/36

Prof Nehal N Shah

ECED, SCET


6

Switching techniques

Circuit switching

Telephone networks

Message switching

Store and forward network

Packet switching

Bandwidth can be managed by splitting data

onto different routes in a busy network.


7/36

Prof Nehal N Shah

ECED, SCET


7

Repeaters

The maximum length for UTP cable in a network,

is 100 meters (approximately 333 feet).

If we need to extend our network beyond that

limit, we must add a device to our network.

This device is called a repeater. The purpose of a repeater is regenerate and

retime network signals at the bit level to allow

them to travel a longer distance on the media.


8/36


9/36

Prof Nehal N Shah

ECED, SCET


9

What is a hub?

The purpose of a hub is to regenerate and retime network

signals.

This is done at the bit level to a large number of hosts (e.g.

4, 8, or even 24) using a process known as concentration.

You will notice that this definition is very similar to the

repeater's, that is why a hub is also known as a multi-port

repeater.


10/36

Prof Nehal N Shah

ECED, SCET


10

Hubs

Active Vs. Passive Hubs

Most modern hubs are active;

they take energy from a

power supply to regeneratenetwork signals.

Some hubs are called passive

devices because they merely

split the signal to multipleusers.

Passive hubs do not

regenerate bits, so they do

not extend a cables length,they only allow two or more

hosts to connect to the same

cable segment.

Intelligent Vs. Dumb

Intelligent hubs have console

ports, which means they can

be programmed to managenetwork traffic.

Dumb hubs simply take an

incoming networking signal

and repeat it to every portwithout the ability to do any

management.


11/36

Prof Nehal N Shah

ECED, SCET


11

Bridges

Bridging occurs at the data link layer, which controls data flow,

handles transmission errors, provides physical addressing, and

manages access to the physical medium.

Bridges provide these functions by using various link layerprotocols that dictate specific flow control, error handling,

addressing, and media access algorithms.

Examples of popular data link layer protocols include Ethernet,

Token Ring, and FDDI. Upper-layer protocol transparency is a primary advantage of

bridging.

Bridges are not required to examine upper-layer information

because they operate at the data link layer or Layer 2 of the OSI

model.

Bridges filter network traffic by only looking at the MAC address,

not protocols.


12/36

Prof Nehal N Shah

ECED, SCET


12

Problems with Bridges When traffic between network segments becomes heavy, bridges can

become a bottleneck and slow down communication.

Bridges always spread and multiply a special kind of data packet.

These data packets occur when a device on a network wants to reachanother device on the network, but does not know the destination address

of the device.

When this occurs, frequently the source sends out a broadcast to all

devices on a network.

Since every device on the network has to pay attention to such

broadcasts, bridges always forward them.

If too many broadcasts are sent out over the network a broadcast storm

can result.

A broadcast storm can cause network time-outs, traffic slowdowns, and

the network to operate at less than acceptable performance.


13/36

Prof Nehal N Shah

ECED, SCET


13

Needs for Faster LAN - I

Network scenario

More users, more networked workstations/computers

Problems

Collisions, low throughput, slow response

Screen Monitor II

LAN

Screen Monitor II

Screen Monitor IIScreen Monitor II


14/36

Prof Nehal N Shah

ECED, SCET


14

Needs for Faster LAN - II

Network scenario Higher desktop processing power

Faster CPU and I/O bus

Problems Bottleneck is found at 10mbps Ethernet / 16mbps token ring

bandwidth

Screen Monitor II

PCI BusEISA Bus

LAN


15/36


16/36

Prof Nehal N Shah

ECED, SCET


16

Key Driving Factors

More network users

More new network application

More powerful CPUs in systems

Increased network loading

Changing organizational structure

Switching is an

affordable andscaleableSolution


17/36


18/36

Prof Nehal N Shah

ECED, SCET


18

Benefits of LAN Switching

Number of Collisions

reduced

Simultaneous multiple

communications

High speed uplinks

Improved network

response Increased user productivity

maximizes the bandwidth

available on the sharedmedium

Most traffic is delivered on

a need to see basis


19/36

Prof Nehal N Shah

ECED, SCET


19

Switching Technology

Multiple 10/100

Mbps conversations

"On-the-fly" packet

switching

Local conversationsare filtered

EthernetEthernetSwitchSwitch


20/36


21/36

Prof Nehal N Shah

ECED, SCET


21

Full Duplex

Bi-directional transmission

Two full 10 or 100 Mbps data paths

Doubles Ethernet capacity to 20 or 200 Mbps

"Collision-free" Ethernet

Supports 10Base-T, 10Base-F, 100Base-TX or 100Base-FX

10Mbps

10Mbps

10Mbps

10Mbps


22/36

Prof Nehal N Shah

ECED, SCET


22

Auto-negotiation

Exchange information

about their abilities over a

link segment

Automatic configuration

Achieve the best possible

duplex mode of operationover a link

Detection of highest speed


23/36

Prof Nehal N Shah

ECED, SCET


23

An Ethernet switch mustexhibit:

Low latency (Tens of

microseconds)Cut-through, or store &

forward technologyMassively parallel switchingScalable architecture

What is a Switch?


24/36

Prof Nehal N Shah

ECED, SCET


24

Store-&-Forward Operation

Receives complete packet

Checks to see that the

packet is not a fragment or

runt resulting from a

collision

Does a CRC to verify

non-corrupted packet

Forward packet based on

forwarding table

If packet is a fragment or

runt or fails the CRC check,

it is discarded

Store Forward

CRC2 3

1 4


25/36

Prof Nehal N Shah

ECED, SCET


25

Store-&-Forward

Pros

Verifies frame checksum before forwarding

Filter runts and check CRCs

Supports mixed speed and LAN ports

Cons

Longer latency (latency=time of frame length

+ processing) More features add delay


26/36


27/36

Prof Nehal N Shah

ECED, SCET


27

Cut-through Operation

Pros

Could be short and consistent latency

Most appropriate in single speed, collision-freeconfiguration

Applicable in single speed workgroup and backbone

switches

Cons

Not capable of detecting and discarding bad packets

Increasing latency w/congestion and w/speed mismatch

Filtering requirements can void cut-through

Must run at the same speed


28/36

Prof Nehal N Shah

ECED, SCET


28

Common Switch Problem

The memory for switch forwarding tables

is expensive

Most of the entries are on the port that

faces the rest of the network


29/36

Prof Nehal N Shah

ECED, SCET


29

Throughput

Throughput = the max. No. of packets per second a switch

can forward without loss across all I/O ports

Normally understood to be in terms of minimum-sized

packets (in Ethernet, 64-byte packets)

Throughput is an aggregated number for a switch

Usually based on implicit/explicit optimal traffic distribution

across all switch ports

Wire-speed = throughput to keep pace with max.

Theoretical frame transmission rates across its I/O ports


30/36

Prof Nehal N Shah

ECED, SCET


30

Buffer Capacity & Congestion Control

Buffer memory allocated to I/O ports; Minimise

frames loss

More than 1 input port is contending for a single output

port

A high-speed input port is forwarding to a single, low-

speed output port

In general, switches with larger buffer capacitywill lose fewer frames during periods of

congestion

In backbones, larger buffer capacity or dynamicbuffer allocation are highly desirable


31/36

Prof Nehal N Shah

ECED, SCET


31

MAC Address Capacity

The number of station addresses that can

be stored in lookup tables

Used for making forwarding decisions

This number establishes the maximum

number of stations that can be included in

a single shared segment attached to a

switch


32/36

Prof Nehal N Shah

ECED, SCET


32

Switching Complementary to Routing

Routers are important

for broadcast filtering,

WAN access, security,and to connect

dissimilar LANs

Ethernet switching

provides throughput

enhancement


33/36


34/36

Prof Nehal N Shah

ECED, SCET


34

What Operates At Which Layer?

Layer 1 Media, Repeaters / Hubs

Layer 2 NIC, Switch / Bridge

Layer 3 Router

Layer 1 to 7 Host / Server


35/36

Prof Nehal N Shah

ECED, SCET


35

Router

The router work at network layer or at layer 3 and make decisions

based on groups (Classes) of network addresses

Routers can also connect different layer 2 technologies, such as

Ethernet, Token-ring, and FDDI. Because of the ability to route packets routers have become the

backbone of the Internet, running the IP protocol.

Routers have both LAN and WAN interfaces.

The two main functions of routers are

selection of best paths for incoming data packets

switching of packets to the proper outgoing interface

Routers accomplish this by building routing tables and exchangingthe network information contained within them with other routers.

You can configure routing tables, but generally they are

maintained dynamically by using a routing protocol


36/36

09 switching devices

Documents