You are on page 1of 115

Chapter 5

LAN Architecture and Protocol

Jun Kyun Choi

jkchoi@icu.ac.kr
Tel) (042) 866-6122

jkchoi@icu.ac.kr 1
Contents
ˆ LAN Switching
ˆ LAN Design
z Network Design Goals, Components, Methodology
z Layer 1 Design
z Layer 2 Design and Layer 3 Design

jkchoi@icu.ac.kr 2
LAN Switching - I

jkchoi@icu.ac.kr 3
Contents
ˆ Effects of switching, bridging, and routing on
network throughput
ˆ Benefits of Ethernet, Fast Ethernet, and VLAN.

jkchoi@icu.ac.kr 4
Introduction
ˆ Network designers are using switches and routers
instead of bridges and hubs.
z Problems in LAN and possible solutions that can
improve performance.
z Effect of LAN congestion on performance.
z Advantages of LAN segmentation.
z Advantages and disadvantages of using bridges,
switches, and routers for LAN segmentation.

jkchoi@icu.ac.kr 5
LAN Communication Problems
ˆ Factors for expanding the capabilities of LANs
z Growing population of network users
z Faster CPUs – workstations with 50 – 75 MIPS are common,
and I/O speeds have increased.
z Faster operating systems – multitasking environment allows
for simultaneous network transactions
z Network-intensive applications – increased use of
client/server applications

jkchoi@icu.ac.kr 6
The Ethernet/802.3 Interface - 1
ˆ Performance of shared-medium Ethernet/802.3
LAN can be affected by
z Data frame broadcast delivery nature
z CSMA/CD allows only one station to transmit at a time.
z Multimedia applications with higher bandwidth demands
z Normal latency as the frames travel across the medium and
networking devices and the latency due to repeaters.
z Extending the distances of LANs with repeaters.

jkchoi@icu.ac.kr 7
The Ethernet/802.3 Interface - 2
ˆ Goal
z To provide a best-effort delivery service.
z Allow all devices on the shared medium to transmit on an
equal basis.
ˆ Inherent problem: collisions
ˆ Half-Duplex Ethernet
z each host checks the network to see whether data is being
transmitted before it transmits data.
z If the network is already in use, the transmission is delayed.

jkchoi@icu.ac.kr 8
Congestion
ˆ There is a need for bandwidth (network capacity)
>> 10 Mbps, because of:
z more powerful computers and network-intensive applications
z increase in the transmission of large graphics file, images,
video
z Increase in the number of network users
ˆ Network congestion occurs
Î slower response times, longer file transfers, users becoming
less productive
z More bandwidth is needed or the available bandwidth must be used more
efficiently.

jkchoi@icu.ac.kr 9
Latency
ˆ Latency (propagation delay) – the time a frame (or
packet) of data takes to travel from the source station to
its final destination on the network.
ˆ Ethernet LANs use CSMA/CD Î there must be latency
to detect collisions and negotiate transmission rights.
ˆ Latency depends on:
z Distance
z Number of devices
z Nature of devices - Routers conduct more complex and time-
consuming decision making than switches

jkchoi@icu.ac.kr 10
Ethernet Transmission Times
ˆ Transmission time is the time it takes a frame (or packet) to
move from the data link layer to the physical layer (physical
cabling).

Packet Size in Bytes Transmission Time in Microseconds

64 51.2
512 410
1000 800
1518 1214

jkchoi@icu.ac.kr 11
Extending Shared-Media LANs with Repeaters - 1

ˆ The distance that a LAN can cover is limited due to


attenuation, which is caused by the resistance in the
medium.
ˆ Repeater is a physical-layer device that boosts (regenerates)
the signal on a LAN.
ˆ Advantages:
z Network can cover a greater distance
z More users can share that same network

jkchoi@icu.ac.kr 12
Extending Shared-Media LANs with Repeaters - 2

ˆ Repeaters increase the collision domain size.


ˆ Repeaters increase the broadcast domain size.
ˆ The broadcast and collision problems are
compounded.
ˆ Repeaters can’t filter traffic.
ˆ Multiport repeater = hub

jkchoi@icu.ac.kr 13
Improving LAN Performance
ˆ Theperformance of a shared-medium LAN
can be improved by:
z Full-duplex Ethernet
z LAN segmentation

jkchoi@icu.ac.kr 14
Full-duplex Ethernet - 1
ˆ Allows the transmission of a packet and the reception
of a different packet at the same time.
ˆ Requires the use of:
z two pairs of wires in the cable
z A switched (point-to-point) connection between nodes
(collision free)
ˆ Can use existing medium as long as it meets min.
Ethernet standards:
z 10BaseT/100BaseTX (100 m)
z 100BaseFX (2 km)

jkchoi@icu.ac.kr 15
Full-duplex Ethernet - 2
ˆ To transmit and receive simultaneously, a dedicated
port is required for each node.
ˆ NICs on both ends need to have full-duplex
capabilities.
ˆ Full-duplex Ethernet Switch creates a direct connection
between the transmit (TX) at one end of the circuit and
the receive (RX) at the other end.
ˆ Collision-free domain is created, because TX and RX
of data occurs on separate circuits.

jkchoi@icu.ac.kr 16
Full-duplex Ethernet - 3
ˆ Half-duplex Ethernet – only uses 50 - 60% of the 10
Mbps bandwidth due to collisions and latency
ˆ Full duplex Ethernet offers 100% of the bandwidth in
both directions, producing a potential 20-Mbps
throughput – 10-Mbps TX and 10-Mbps RX

jkchoi@icu.ac.kr 17
LAN Segmentation - 1
ˆ A network can be divided into smaller units called segments,
each of which uses CSMA/CD and maintains traffic between
users on the segment.
ˆ By using segments, fewer users/devices are sharing the same
10 Mbps when communicating within the segment.
ˆ Each segment is its own collision domain.
ˆ Segmentation can decrease network congestion within each
segment.
ˆ Data passed between segments is transmitted on the
backbone of the network using a bridge, router, or switch.

jkchoi@icu.ac.kr 18
LAN Segmentation - 2
ˆ Segmentation with bridges
ˆ Segmentation with routers
ˆ Segmentation with switches

jkchoi@icu.ac.kr 19
Segmentation with Bridges
ˆ Bridges “learn” a network’s segmentation by building
address tables that contain the MAC address of each
network device and which interface to use to reach that
device.
ˆ They are Layer 2 devices.
ˆ They are transparent to the other devices.
ˆ Increase the latency in a network by 10 - 30%, due to the
decision making required of the bridges in transmission:
z A bridge is a store-and-forward device because it must examine the
destination addr field and calculate the CRC before forwarding the frame.
ˆ If the destination port is busy, the bridge can temporarily
store the frame until the port is available.

jkchoi@icu.ac.kr 20
Segmentation with Routers
ˆ Routers are more advanced than bridges
ˆ Operates at the network layer.
ˆ Creates the highest level of segmentation by
forwarding data to the hub, to which w/s are
connected.
ˆ Makes forwarding decisions by examining the
destination addr. on the packet and looking in its
routing table.
ˆ Must examine a packet to determine the best path
for packet forwarding – takes time.
jkchoi@icu.ac.kr 21
Segmentation with Routers
ˆ Protocols that require an acknowledgement from
the receiver for every packet have a 30 - 40% loss of
throughput.
ˆ Protocols that require minimal acknowledgements
(e.g. sliding-window protocols) suffer a 20 - 30%
loss of throughput.

jkchoi@icu.ac.kr 22
Segmentation with Switches - 1
ˆ A switch can segment a LAN into microsegments
(single-host segments), creating collision-free
domains from one larger collision domain.
ˆ All hosts connected to the switch are still in the
same broadcast domain.
ˆ A LAN switch is a very high-speed multiport bridge
with one port for each node (or LAN segment)
ˆ Make frame forwarding decisions by building a
table of the MAC addresses of the hosts attached to
each port.

jkchoi@icu.ac.kr 23
Segmentation with Switches - 2
ˆ A computer connected directly to a switch is its own
collision domain and accesses the full 10 Mbps.
ˆ A switched LAN behaves as though it has only two
nodes: the sending node and the receiving node,
which share the 10-Mbps bandwidth between them.
ˆ Switched LAN provide more throughput than
LANs connected by bridges or hubs.
ˆ Available bandwidth can reach close to 100%.

jkchoi@icu.ac.kr 24
Segmentation with Switches - 3
ˆ Increases the bandwidth available on a network by
creating dedicated network segments (point-to-
point connections) and connecting those segments
in a virtual network within the switch.
ˆ The virtual network circuit exists only when two
nodes need to communicate (i.e., virtual circuit)
ˆ Drawback: switches cost more than hubs.

jkchoi@icu.ac.kr 25
Bridging

(Note) FDB: Forwarding Database

jkchoi@icu.ac.kr 26
LAN Switching - II

jkchoi@icu.ac.kr 27
Contents
ˆ Switchingand VLANs.
ˆ Spanning Tree Protocol

jkchoi@icu.ac.kr 28
Switching and VLANs
ˆ Switching – decreases congestion in LANs by
reducing traffic and increasing bandwidth.
ˆ Switches can be installed without disrupting
existing network traffic – replace hubs.
ˆ Two basic operations of switching equipment
z Switching data frames – when a frame arrives on an input
media and is transmitted to an output media.
z Maintaining switching operations – switches build and
maintain switching tables.

jkchoi@icu.ac.kr 29
Bridging - 1
ˆ A bridge connects two or more LAN segments, and transmit
frames from one segment to other segment(s).
ˆ A bridge begins by examining the MAC address of the
incoming frame and builds a table of known destinations.
ˆ If destination of frame is on the same segment as the source
of frame,
z then bridge drops the frame
z else bridge transmits the frame on destination segment only
ˆ If bridge does not know the destination segment
z bridge transmits frame on all segments except the source segment
(flooding)
ˆ Benefit: limits traffic to certain network segments.

jkchoi@icu.ac.kr 30
Bridging - 2
ˆ Both bridges and switches connect LAN segments,
use a table of MAC addresses to determine the dest.
segment, and therefore reduce traffic.
ˆ Switches operate at much higher speeds than
bridges and can support new functionality, (e.g.,
virtual LANs, VLANs).
ˆ Bridges switch using software; switches switch
using hardware.

jkchoi@icu.ac.kr 31
LAN Switch Latency
ˆ Each switch adds latency to a network
z e.g, a switch between a workstation (w/s) and a server adds
21 µs to the transmission process.
z Total transmission time of a 1000-byte packet = 800 + 21 =
821 µs (assuming a transmission rate of 10 Mbps).
ˆ Type of switching used, e.g., cut-through, can help
overcome the latency.

jkchoi@icu.ac.kr 32
Layer 2 and Layer 3 Switching
ˆ Two methods of switching:
z Layer 2 switching – used by (Layer 2) switches to
forward frames based on destination MAC address.
z Layer 3 switching – used by routers to forward
(route) a packet based on network-layer destination
address.

jkchoi@icu.ac.kr 33
Layer 2 Switching
ˆ Layer 2 switching builds and maintains a switching table
that keeps track of which MAC addresses belong to each
port (interface).
ˆ If Layer 2 switch does not know which port to send the
frame, it broadcasts the frame out all its ports. When a
reply is returned, the switch learns the location of the new
address and updates the switching table.
ˆ Layer 2 addresses are fixed with a device and are
determined by the manufacturer. They are derived in two
parts: the manufacturing (MFG) code and the unique
identifier.

jkchoi@icu.ac.kr 34
Layer 3 Switching - 1
ˆ Operates at the network layer and supports router
functionality.
ˆ Layer 3 addresses are determined by the network
administrator. By creating Layer 3 addresses, he
creates local areas that act as single addressing
units (like streets, districts, etc.) and assigns a
number to each local entity.
ˆ Addressing is used by IP, IPX, and AppleTalk.

jkchoi@icu.ac.kr 35
Layer 3 Switching - 2
ˆ Routers can adheres to and create a hierarchical addressing
structure. Thus, a routed network can tie a logical
addressing structure to a physical infrastructure, e.g.,
through TCP/IP subnets for each network segment.
ˆ Traffic flow in a switched (flat) network is therefore
inherently different from that in a routed (hierarchical)
network.
z Hierarchical networks offer more flexible traffic flow than
flat networks because hierarchical networks can use the
hierarchy to:
z determine optimal paths
z contain broadcast domains

jkchoi@icu.ac.kr 36
Implications of Layer 2 and Layer 3 Switching - 1
ˆ Increased need for greater bandwidth in traditional shared-
media environments Æ network designers are replacing
hubs in wiring closets with switches.
ˆ Layer 2 switches use micro segmentation to satisfy the
demands for more bandwidth, but there is increasing
demands for intersubnet communication. E.g., every time a
user accesses resources located on different subnets, the
traffic must go through a Layer 3 device.
ˆ To avoid bottleneck, Layer 3 capabilities can be added
throughout the network, alleviating the burden on
centralized routers.

jkchoi@icu.ac.kr 37
Implications of Layer 2 and Layer 3 Switching - 2

ˆ A switch improves bandwidth by:


z separating collision domains
z selectively forwarding traffic to the appropriate network
segments

jkchoi@icu.ac.kr 38
How a Switch Learns Addresses
ˆ A switch can learn the address of each device on the
network by reading the source address of each packet
transmitted and noting the port where the frame entered the
switch.
ˆ Switch updates its forwarding database.
ˆ Addresses are learned dynamically – as new addresses are
read, they are learned and stored in content-addressable
memory (CAM).
ˆ Aged or old addresses are removed:
z When an address is stored, it is time stamped. Each time an
address is referenced or found in CAM, it receives a new
time stamp.

jkchoi@icu.ac.kr 39
The Benefits of Switching
ˆ Allows many users to communicate in parallel with
virtual circuits and dedicated network segments in
a collision-free environment – maximizes the
bandwidth available on the shared medium.
ˆ Cost-effective, because existing hardware and
cabling can be reused.
ˆ When combined with the software to configure
LANs Æ great flexibility in network management

jkchoi@icu.ac.kr 40
Symmetric and Asymmetric Switching - 1
ˆ Characterizing based on the bandwidth allocated to
each switch port:
z Symmetric Switching
z Asymmetric Switching
ˆ Symmetric switch provides switched connections
between ports with the same bandwidth
z e.g., all 10-Mbps ports.

jkchoi@icu.ac.kr 41
Symmetric and Asymmetric Switching - 2
ˆ Asymmetric switch provides switched connections between
ports of unlike bandwidth, e.g., a combination of 10-Mbps
and 100-Mbps ports.
ˆ Asymmetric switching makes the most of client/server
network traffic flows where multiple clients are
communicating with a server at the same time, requiring
more bandwidth dedicated to the switch port that the server
is connected to Æ prevent a bottleneck at that port.
ˆ Memory buffering is required to allow traffic from the 100-
Mbps port to be sent to a 10-Mbps port without causing too
much congestion.

jkchoi@icu.ac.kr 42
Memory Buffering
ˆ May be used:
z to store and forward packets to the correct port(s) or
z When the destination port is busy.
ˆ Memory buffer – the area of memory where the
switch stores the destination and transmission data.
Can use two methods:
z Port-based memory buffering
z Shared memory buffering

jkchoi@icu.ac.kr 43
Switching Methods
ˆ Two switching modes for frame forwarding
z Store-and-forward - the entire frame is received before any
forwarding takes place.
z Latency occurs while the frame is being received

z High error detection, because of the time available for


error checking while waiting for the entire frame to be
received.
z Cut-through – the frame is forwarded after the switch reads
the dest. addr., but before the entire frame arrives.
z Decreases the latency, but poor error detection

jkchoi@icu.ac.kr 44
Switching Methods (Cont’d)
ˆ Cut-through – two forms
z Fast-forward switching
z Offers the lowest level of latency by immediately
forwarding a packet after receiving the dest. addr.
z Fragment-free switching
z Filters out collision fragments (< 64 bytes), which are the
majority of packet errors, before forwarding begins.

jkchoi@icu.ac.kr 45
Virtual LANs (VLANs)
ˆ An Ethernet switch physically segments a LAN into collision
domains, but each segment is still part of one broadcast
domain Æ all nodes on all segments can see a broadcast
from a node on one segment.
ˆ A VLAN is a logical grouping of network devices (or users)
that are not restricted to a switched segment.
ˆ The devices can be grouped by function, department, etc.,
regardless of their physical segment location.
ˆ A VLAN creates a single broadcast domain that is not
restricted to a physical segment and is treated like a subnet.

jkchoi@icu.ac.kr 46
Spanning Tree Protocol (STP) - 1
ˆ Main function of STP is to allow duplicate paths in a
switched/bridged network without incurring the latency
effects of loops in the network.
ˆ Bridges and switches make forwarding decisions (for unicast
frames) based on the destination MAC addr.
ˆ If the MAC addr is unknown, the device floods the frame
out all ports (also true for broadcast frames).
ˆ The Spanning Tree Algorithm prevents loops by calculating
a stable spanning-tree network topology.

jkchoi@icu.ac.kr 47
Spanning Tree Protocol (STP) - 2
ˆ When creating fault-tolerant networks, a loop-free path
must exist between all network nodes.
ˆ STP frames (called Bridge Protocol Data Units – BPDU) are
sent and received by all switches in the network at regular
intervals and are used to determine the spanning-tree
topology.
ˆ STP is used on all Ethernet - and Fast Ethernet-based
VLANs.
ˆ STP detects loops and breaks them by placing some
connections in a standby mode, which are activated in the
event of a connection failure.
ˆ A separate instance of STP runs within each configured
VLAN.
jkchoi@icu.ac.kr 48
Spanning Tree Protocol States - 1
ˆ The state for each VLAN is initially set by
configuration according to policy and later
modified by the STP process based on traffic
patterns and potential loops.
ˆ STP states:
z Blocking – No frames forwarded, BPDUs heard
z Listening – No frames forwarded, listening for data frames
z Learning – No frames forwarded, learning addresses
z Forwarding – Frames forwarded, learning addresses
z Disabled – No frames forwarded, no BPDUs heard

jkchoi@icu.ac.kr 49
Spanning Tree Protocol States - 2
ˆ show spantree command – to determine the status,
cost, and priority of ports and VLANs.
ˆ After the port-to-VLAN state is set, STP determines
whether the port forwards or blocks frames.
ˆ Ports can be configured (Port Fast mode) to
immediately enter STP forwarding state when a
connection is made instead of following the usual
sequence (blocking Æ listening Æ learning Æ
forwarding) – useful for immediate access to a
server.
jkchoi@icu.ac.kr 50
LAN Design –
Network Design Goals, Components, Methodology,
and Layer 1 Design

jkchoi@icu.ac.kr 51
Introduction
ˆ Trend: Increasingly complex environments involving
multimedia (or multiple media types) and interconnection to
networks outside a controlled LAN.
ˆ Network design is a critical step to ensure a fast and stable
network.
ˆ If a network is not design properly:
z Many unforeseen problems can arise
z Network growth can be jeopardized
ˆ An overview of the LAN design process:
z LAN Design Goals and Components
z Network Design Methodology
z Layer 1 Design

jkchoi@icu.ac.kr 52
LAN Design Goals
ˆ The first step in designing a LAN is to establish and
document the design goals, which are particular to each
organization and situation.
ˆ For most network designs, the following are required:
z Functionality: the network must allow users to meet their job
requirements. The network must provide connectivity with reasonable
speed and reliability.
z Scalability: the network must be able to grow. That is, the initial design
should grow without any major changes to the overall design.
z Adaptability: the network must be designed with an eye toward future
technologies, and it should include no element that would limit
implementation of new technologies as they become available.
z Manageability: the network should be designed to facilitate network
monitoring and management to ensure ongoing stability of operation.

jkchoi@icu.ac.kr 53
Network Design Components
ˆ To design LANs for high-speed technologies and
multimedia-based applications, the following
critical components of LAN design should be
addressed:
z The function and placement of servers
z Collision detection
z Segmentation
z Bandwidth vs. broadcast domains

jkchoi@icu.ac.kr 54
Function and Placement of Servers
ˆ Servers provide file sharing, printing,
communication, and application services.
ˆ Typically do not function as workstations.
ˆ Two distinct classes:
z Enterprise servers
z Supports all network users by offering services, e.g., e-mail or DNS,
which everyone in an organization would need because it is a
centralized function.
z Should be placed in the main distribution facility (MDF)
z Traffic to the servers does not need to be transmitted across other
networks.

jkchoi@icu.ac.kr 55
Function and Placement of Servers (Cont’d)
z Workgroup servers
z Supports a specific set of users, offering services which only a few
groups of people would need.
z Should be placed in the intermediate distribution facilities (IDFs)
closest to the users accessing the applications on these servers.
z Traffic only has to travel the network infrastructure to that IDF,
ˆ Within the MDF and IDFs, the Layer 2 LAN switches
should have 100 Mbps or more allocated for these servers.

jkchoi@icu.ac.kr 56
Intranets
ˆ Intranets are designed to be accessed by users who
have access privileges to an organization’s internal
LAN.
ˆ Within an intranet, Web servers are installed, and
browser technology is used as the common front
end.
ˆ Intranet can cause an increase in needed network
bandwidth.
ˆ Bandwidth added to backbone Î new desktops and
servers should be outfitted with 10/100-Mbps NICs
to provide configuration flexibility.

jkchoi@icu.ac.kr 57
Collision Detection - 1
ˆ To decrease collision detection and media
contention on a network, careful selection and
placement of networking devices is needed.
ˆ Contention – excessive collisions on Ethernet
caused by too many devices.
ˆ Too many broadcasts when there are:
z Too many client packets
z Too many server packets
z Too many routing table updates
z Too many broadcasts due to protocols (e.g., ARP)
ˆ Contention media access does not scale (or allow for
growth) – Ethernet’s disadvantage.
jkchoi@icu.ac.kr 58
Collision Detection - 2
ˆ As traffic increases on the shared media, rate of
collisions also increases.
ˆ An excessive number of collisions reduces actual
available bandwidth – to about 35% - 40% of the
full 10 Mbps.
ˆ Reduction in bandwidth can be remedied by
segmentation with bridges, switches, or routers.

jkchoi@icu.ac.kr 59
Segmentation
ˆ Segmentation – the process of splitting a single collision
domain into two or more collision domains.
z e.g. Layer 2 bridges or switches can be used to segment a logical bus
topology and create separate collision domains, resulting in more
bandwidth being available to individual stations.
ˆ The scalability of the bandwidth domain depends on the
total amount of traffic.
ˆ The scalability of the broadcast domain depends on the total
broadcast of the traffic.
ˆ The entire bus topology still represents a single broadcast
domain, because bridges and switches forward broadcast
(FF-FF-FF-FF-FF-FF) frames.
jkchoi@icu.ac.kr 60
jkchoi@icu.ac.kr 61
Bandwidth vs Broadcast Domains
ˆ Bandwidth domain – everything associated with
one port on a bridge or switch.
z For Ethernet switch, bandwidth domain = collision domain.
ˆ All workstations within one bandwidth domain
compete for the same LAN bandwidth resource.
ˆ All traffic from any host in the bandwidth domain
is visible to all the other hosts.
ˆ For Ethernet, two stations can transmit at the same
time, causing a collision Î collision domain.

jkchoi@icu.ac.kr 62
Network Design Methodology
ˆ A LAN should be designed and implemented
according to the following steps:
1. Gathering the users’ requirements and expectations
2. Analyzing the requirements
3. Designing the Layer 1,2, and 3 LAN structure (i.e.,
topology)
4. Documenting the logical and physical network
implementation

jkchoi@icu.ac.kr 63
Gathering Requirements - 1
ˆ Gather information about the organizational structure:
z History and current status
z Projected growth
z Operating policies and management procedures
z Office systems and procedures
z Viewpoints of the LAN users
ˆ To determine how much training will be required and how many
people will be needed to support the LAN, we need to know:
z Who are the prospective network users?
z What is their skill level?
z what are their attitudes toward computers and computer applications?

jkchoi@icu.ac.kr 64
Gathering Requirements - 2
ˆ Are there documented policies in place?
z Has some data been declared mission critical?
z Have some operations been declared mission critical?
z (Mission-critical data and operation are those that are considered key
to business)
z What network protocols are allowed?
z Are only certain desktop hosts supported?
ˆ Who in the organization has authority over addressing,
naming, topology design and configuration?
z Central MIS department that controls everything, or
z Small MIS department – authority delegated to departments

jkchoi@icu.ac.kr 65
Gathering Requirements - 3
ˆ Identify the organization resources and constraints.
Two categories of organization resources:
z Computer hardware/software
z Must be documented
z Organization’s projected needs must be identified
z Human resources
ˆ How are these resources currently linked and shared?
ˆ What financial resources are available?
ˆ Documenting these helps us estimate cost and develop a
budget for the LAN
ˆ Understand performance issues of any existing
network.

jkchoi@icu.ac.kr 66
Analyzing Requirements
ˆ Analyze the requirements of the network and its users.
z Network user needs constantly change, e.g., bandwidth
requirement increases as more multimedia applications
become available
ˆ Assessing user (information) requirements
z Need to supply prompt and accurate information to LAN
users
ˆ Availability and network traffic. Availability measures the
usefulness of the network, and is affected by:
z Throughput, Response time, Access to resources
ˆ Network design seeks to provide the greatest availability for
the least cost.

jkchoi@icu.ac.kr 67
Designing the Network Topology
ˆ To decide on an overall LAN topology that will
satisfy the user requirements.
ˆ Concentrates on star topology and extended star
topology.
z Uses Ethernet 802.3 CSMA/CD technology
z By far the dominant configuration in industry
ˆ LAN topology design can be broken into:
z Physical layer
z Data Link layer
z Network layer
jkchoi@icu.ac.kr 68
Designing the Layer 1 Topology
ˆ Cabling
ˆ Star topology
ˆ Extended star topology

jkchoi@icu.ac.kr 69
Cabling
ˆ Designs issues:
z Type of cabling
z Category 5 UTP – used in the horizontal runs
z Fiber-optic cable – used in the backbone and risers
z EIA/TIA 568 standard for layout and connection of wiring schemes
z Overall structure
ˆ Distance limitations
ˆ Evaluate strengths and weaknesses of various
topologies - a network is only as effective as its
underlying cable.
ˆ Cable upgrade should take priority over any other
needed changes.
jkchoi@icu.ac.kr 70
Cabling (Cont’d)
ˆ Cabling systems should conform to industrial
stands (e.g. EIA/TIA 568)
ˆ EIA/TIA 568 standard specifies:
z Every device connected to the network should be linked to a
central location with horizontal cabling (provided all the
hosts concerned are < 100m (distance limitation for CAT 5
UTP Ethernet)(EIA/TIA 568B standards)

jkchoi@icu.ac.kr 71
jkchoi@icu.ac.kr 72
Star Topology
ˆ With only one wiring closet
ˆ MDF includes one or more horizontal cross-
connect (HCC) patch panels, which are used
to connect the Layer 1 horizontal cabling
with the Layer 2 LAN switch ports.
ˆ Uplink port of switch (unlike other ports
because it does not cross over) is connected to
the Ethernet port of Layer 3 router using
patch cable.

jkchoi@icu.ac.kr 73
Extended Star Topology - 1
ˆ When hosts are outside the 100-meter limitation for
Cat 5 UTP Ethernet, more than one wiring closet
are needed Î multiple catchment areas.
ˆ TIA/EIA 568-A standard specifies that IDFs
(secondary wiring closets) should be connected to
the MDF by using vertical cabling (backbone
cabling)
ˆ Vertical cross-connect (VCC) is used to interconnect
IDFs to MDF.
ˆ Vertical cable length typically > 100 m Î fiber-
optic cabling is used

jkchoi@icu.ac.kr 74
Extended Star Topology - 2
ˆ Fast Ethernet
z Ethernet upgraded to 100 Mbps
z Uses broadcast-oriented logical bus topology of 10BaseT
z Uses CSMA/CD for Media Access Control
z With several standards based on:
z Copper-pair wire (100BaseTX)
z Fiber-optic cable (100BaseFX)
z Used to connect the MDF to IDF

jkchoi@icu.ac.kr 75
jkchoi@icu.ac.kr 76
jkchoi@icu.ac.kr 77
Layer 1 Documentation - Logical Diagram
ˆ Network topology model without all the detail of exact
installation path of cabling.
ˆ A snapshot view of the LAN implementations
ˆ Useful in troubleshooting problems and implementing
expansion in the future
ˆ Elements include:
z Exact locations of the MDF and IDF wiring closets
z Type and quantity of cabling used to interconnect IDFs with
MDF, along with number of spare cables available for
increasing the bandwidth between the wiring closets.
z Detail documentation of all cable runs, the identification
nos., and which port on the HCC or VCC the run is
terminated on.
jkchoi@icu.ac.kr 78
Appendix: Fast Ethernet (IEEE 802.3u)
ˆ Aim: To obtain an order of magnitude increase in speed
over 10BaseT Ethernet, while retaining the same wiring
systems, MAC method, and frame formats.
ˆ IEEE 802.3 spec. allows for a total cable length (with
repeaters) of 2.5km.
z Worst-case signal propagation delay = Time for the signal to propagate
twice the max. length.
z The standard allows a worst-case signal propagation delay (including
repeater delay) of 50 µs which is equivalent to 500 bits at 10 Mbps. A
safety margin is added to give a minimum frame size of 512 bits (64
bytes).
ˆ If this max. length is reduced, then CSMA/CD can be used
with higher bit rates.
jkchoi@icu.ac.kr 79
Appendix: Fast Ethernet (IEEE 802.3u)
ˆ Most 10BaseT installations use < 100 m of cable between
station and hub.
⇒ Max. distance between any two stations is 200 m
⇒ Worst-case path length for collision detection = 400 m
⇒ A higher bit rate can be used while still retaining the same CSMA/CD
MAC method and min. frame size.
ˆ In the standard, the data rate is set at 100 Mbps. Hence,
100BaseT.
ˆ The committee defined a number of options to be used with
different transmission media.
ˆ Uses the same IEEE802.3 MAC protocol (CSMA/CD) and
frame format.
ˆ Topology is similar to that of 10BaseT, viz, star-wire
topology.

jkchoi@icu.ac.kr 80
100BASE-T Options - 1
ˆ 100BASE-X - use two physical links between
nodes
z 100BASE-TX
z 2 Category 5 UTP data-grade wiring is required to support high
data rate.
z 2 STP
z 100BASE-FX
z 2 Optical Fiber
ˆ 100BASE-T4
z 4 Category 3 or Category 5 UTP

jkchoi@icu.ac.kr 81
100BASE-T Options - 2

jkchoi@icu.ac.kr 82
100BASE-X
ˆ Unidirectional data rate of 100 Mbps is achieved
over a single link (Single TP or single optical
fiber)
ˆ 2 physical medium spec.
z 100BASE-TX - two pairs of TP (STP or Cat 5 UTP)
cable (one for transmission, one for reception)
z 100BASE-FX - two optical fiber cables (one for
transmission, one for reception)

jkchoi@icu.ac.kr 83
100BASE-T4
ˆ In many building, 100BASE-X options requires the
installation of new cable
ˆ 100BASE-T4 is designed to produce a 100-Mbps data rate
over lower-quality (voice-grade) Category 3 cable, taking
advantage of large installed base of Cat 3 cable in office
bldg. (The use of Cat 5 cable is optional.)
ˆ Unlikely for Cat 3 cable to achieve 100 Mbps on a single TP.
ˆ 100BASE-T4 specifies that data stream is split into 3
separate streams, each with an effective data rate of 100/3
Mbps.
ˆ 4 TP are used - 3 pairs for transmission and 3 pairs for
reception
z Two of the pairs must be configured for bidirectional transmission.

jkchoi@icu.ac.kr 84
LAN Design –
Layer 2 Design and Layer 3 Design

jkchoi@icu.ac.kr 85
Introduction
ˆ Layer 2 Design
ˆ Layer 3 Design

jkchoi@icu.ac.kr 86
Designing the Layer 2 LAN Topology - 1
ˆ Purpose of Layer 2 devices:
z Flow control
z Error detection
z Error correction
z Reduce network congestion
ˆ Most common Layer 2 devices (other than NIC)
z Bridges
z LAN switches
ˆ Layer 2 Devices determines the size of collision domains and
broadcast domains.
ˆ We’ll concentrate on the implementation of LAN switching
at Layer 2.

jkchoi@icu.ac.kr 87
Designing the Layer 2 LAN Topology - 2
ˆ Collisions and collision domain size negatively
affect network performance.
ˆ LAN switching – micro segment the network Æ
eliminating collisions and reducing the size of
collision domains.
ˆ Characteristic of LAN switch – bandwidth
allocation on a per-port basis (asymmetric
switching):
z Allowing more bandwidth to vertical cabling, uplinks, and
servers.
z Provides switched connections between ports of unlike
bandwidth, e.g., a combination of 10-Mbps and 100-Mbps
ports.
jkchoi@icu.ac.kr 88
Effect of Micro Segmentation on a Network
ˆ Micro Segmentation – using bridges (CNAP?) and
switches to boost performance for a workgroup or a
backbone.
ˆ Switches can be used with hubs to provide the
appropriate level of performance for different users
and servers.

jkchoi@icu.ac.kr 89
Determining the Number of Cable Runs and
Drops - 1
ˆ By installing:
z LAN switching at MDF and IDFs
z Vertical cable between MDF and IDFs
z Vertical cable is carrying all the data traffic between MDF
and IDFs Æ its capacity must be > that of the runs between
IDFs and w/s’s
ˆ Horizontal cable runs use CAT 5 UTP (should be
<= 100 m), allowing links at 10 Mbps or 100 Mbps
z Normally, 10 Mbps is adequate

jkchoi@icu.ac.kr 90
Determining the Number of Cable Runs and
Drops - 2
ˆ Asymmetric LAN switches Æ need to determine the
no. of 10-Mbps and 100-Mbps ports needed in MDF
and every IDF, basing on user requirements:
z No. of horizontal cable drops per room
z No. of drops total in any catchment area
z Catchment area – an area that can be served by an internetworking
device, e.g., a hub
z No. of vertical cable runs
ˆ E.g. User requirement:
z 4 horizontal cable runs to be installed to each room
z IDF serving a catchment area covers 18 rooms
z Thus, 4 drops x 18 rooms = 72 switch ports

jkchoi@icu.ac.kr 91
jkchoi@icu.ac.kr 92
Determining the Size of Collision Domains in Hubbed
and Switched Networks

ˆ To determine the size of a collision domain, we


must determine the no. of hosts connected to any
single switch port.
z Also affects how much bandwidth is available to any host.
ˆ One way to implement LAN switching is to installed
shared LAN hubs on the switch ports and connect
multiple hosts to a single switch port
z all hosts connected to the shared LAN hub share the same
collision domain and bandwidth

jkchoi@icu.ac.kr 93
jkchoi@icu.ac.kr 94
jkchoi@icu.ac.kr 95
Diagramming Hub Placement in A Standards-based
Extended Star Topology
ˆ Shared media hubs are generally used in a LAN
switch environment to create more connection
points at the end of horizontal cable runs.
ˆ Should make sure that:
z Design specifications are adhered to
z Bandwidth requirements are met
z Collision domains are kept to a minimum

jkchoi@icu.ac.kr 96
jkchoi@icu.ac.kr 97
Migrating a network from 10 Mbps to 100
Mbps
ˆ Capacity of vertical cabling (between MDF and IDFs) can
be doubled by
z connecting unused fiber optics from the VCC to 100-Mbps
ports on the switch.
ˆ Bandwidth of horizontal cabling can be increased by a
factor of 10 by
z Repatching from the HCC to a 100-Mbps port on the switch
and
z Changing from a 10-Mbps hub to 100-Mbps hub
ˆ It is important to:
z Make sure there are enough 100-Mbps ports to allow for
migration.
z Document the speed at which each active cable drop is running.
jkchoi@icu.ac.kr 98
jkchoi@icu.ac.kr 99
Layer 3 Design
ˆ Layer 3 (network layer) devices, e.g., routers, allow
for:
z Creation of LAN segments
z Communication between segments based on Layer 3
addressing (e.g., IP addressing)
z Segmentation of LAN into physical and logical networks
z Connectivity to WANs, e.g., Internet

jkchoi@icu.ac.kr 100
Layer 3 Router Implementation
ˆ Layer 3 routing
z Addresses the need for physically separate subnets.
z Determines traffic flow between physical segments
based on Layer 3 addressing, e.g., IP network and
subnet.
ˆ Router
z Forwards data packets based on destination
addresses.
z Does not forward broadcasts (e.g., ARP requests)
z Router interface is the entry and exit point of a
broadcast domain
jkchoi@icu.ac.kr 101
jkchoi@icu.ac.kr 102
VLAN Implementation
ˆ VLANs
z Can limit broadcast traffic to within a VLAN and thus create smaller
broadcast domains
z Can provide security by creating the VLAN groups according to function
ˆ Physical port association is used to implement VLAN
assignment.
ˆ Communication between VLANs can occur only through a
router.
z Limits the size of the broadcast domains
z Router determines whether VLAN 1 and talk to VLAN 2 (i.e., a security
scheme can be created based on VLAN assignment.)

jkchoi@icu.ac.kr 103
jkchoi@icu.ac.kr 104
Router provides communication between
VLAN 1 and VLAN 2

jkchoi@icu.ac.kr 105
Using Routers to Create Scalable Networks - 1
ˆ Routers provide scalability because
z They can serve as firewalls for broadcasts.
z They divide networks and subnets, thus, adding structure to
Layer 3 addresses.
z Final step is to develop and document the IP addressing scheme to be
used.
ˆ Routing filters data-link broadcasts and multicasts
ˆ Network protocol addressing and routing provide
built-in scaling.
z By adding router ports with additional subnet or network
addresses (i.e., network-layer addresses), the internetwork
can be segmented as required.

jkchoi@icu.ac.kr 106
Using Routers to Create Scalable Networks - 2
ˆ Whether to use routers or switches?
z Need to ask: “What problem am I trying to solve?”
z If problem is protocol related rather than contention oriented
Æ routers
ˆ Routers can solve these problems:
z Excessive broadcasts
z Protocol that do not scale well
z Security issues
z Network-layer addressing
ˆ Disadvantages of routers:
z More expensive
z Harder to configure than switches
jkchoi@icu.ac.kr 107
A router provides structure to a network by dividing
networks and subnets.

jkchoi@icu.ac.kr 108
Using Routers to Impose Logical Structure
ˆ Routers can be used to provide IP subnets or add structure to
addresses.
ˆ With routers, hosts using protocols with network-layer addressing
can find other hosts without flooding, as used in bridges and
switches.
z If destination address is local
z Sending host can encapsulate the packet in a data-link header and send a
unicast frame directly to the station.
z Router does not see the frame.
z Sending host might use ARP, causing a broadcast, but the broadcast is only a
local one and is not forwarded by router.
z If destination address is not local
z Sending host transmits the packet to router, which sends the frame to the
destination or to the next hop, based on its routing table

jkchoi@icu.ac.kr 109
Using a Router for Segmentation
ˆ The following figure shows an implementation with multiple
physical networks.
ˆ All data traffic from Network 1 destined for Network 2 has
to go through the router.
ˆ There are two broadcast domains.
ˆ The two networks have unique Layer 3 IP addressing and
network/subnetwork addressing schemes.
ˆ In structured Layer 1 wiring scheme, multiple physical
networks are created simply by patching the horizontal and
vertical cabling into a Layer 2 switch using patch cables.
ˆ This implementation:
z Provides for robust security
z Router is the central point for traffic destination

jkchoi@icu.ac.kr 110
jkchoi@icu.ac.kr 111
Documenting the Logical and Physical Network
Implementation
ˆ After the IP addressing scheme is developed, it
should be documented by site and by network with
the site.
ˆ A standard convention should be set for addressing
important network hosts.
ˆ Address scheme should be kept consistent
throughout the entire network.
ˆ By creating addressing maps, we can get a snapshot
of the network.
ˆ Creating physical maps helps us troubleshoot the
network.

jkchoi@icu.ac.kr 112
Address scheme should be kept consistent
throughout the entire network.

jkchoi@icu.ac.kr 113
By creating addressing maps, we can get a
snapshot of the network

jkchoi@icu.ac.kr 114
Physical maps indicate where MDFs and IDFs are
located and where a host is connected to the network.

jkchoi@icu.ac.kr 115

You might also like