Tài liệu Module 5: Designing a Highly Available Physical Network docx

Contents
Overview 1
Lesson: Selecting Networks Based on
Bandwidth Requirements 2
Lesson: Selecting Redundant Paths for
Network Connectivity 14
Lesson: Selecting the Appropriate Network
Topology 20
Lesson: Selecting a Highly Available Switch
Configuration 32
Lesson: Selecting a Highly Available Routing
Configuration 46
Lab: Designing a Highly Available Physical
Network 59

Module 5: Designing a
Highly Available
Physical Network



Information in this document, including URL and other Internet Web site references, is subject to
change without notice. Unless otherwise noted, the example companies, organizations, products,
domain names, e-mail addresses, logos, people, places, and events depicted herein are fictitious,
and no association with any real company, organization, product, domain name, e-mail address,

logo, person, places or events is intended or should be inferred. Complying with all applicable
copyright laws is the responsibility of the user. Without limiting the rights under copyright, no
part of this document may be reproduced, stored in or introduced into a retrieval system, or
transmitted in any form or by any means (electronic, mechanical, photocopying, recording, or
otherwise), or for any purpose, without the express written permission of Microsoft Corporation.

Microsoft may have patents, patent applications, trademarks, copyrights, or other intellectual
property rights covering subject matter in this document. Except as expressly provided in any
written license agreement from Microsoft, the furnishing of this document does not give you any
license to these patents, trademarks, copyrights, or other intellectual property.

 2001 Microsoft Corporation. All rights reserved.

Microsoft, MS-DOS, Windows, Windows NT, Active Directory, BackOffice, FrontPage, Outlook,
PowerPoint, Visio, Visual Studio, Win32, and Windows Media are either registered trademarks or
trademarks of Microsoft Corporation in the U.S.A. and/or other countries.

The names of actual companies and products mentioned herein may be the trademarks of their
respective owners.


iii


Instructor Notes
This module provides students with the knowledge and skills needed to design a
highly available physical network. The information in this module introduces
the physical network of the Web infrastructure and the factors that impact
availability.
After completing this module, students will be able to:

!
Select a network based on bandwidth requirements.
!
Select redundant paths for network connectivity.
!
Select appropriate networking topology for the physical network.
!
Select a highly available switch configuration for the physical network.
!
Select a highly available routing configuration for the physical network.
!
Select the appropriate technologies, configurations, and hardware
components required for designing a highly available physical network.

To teach this module, you need the following materials:
!
Microsoft
®
PowerPoint
®
file 2088A_05.ppt
!
Delivery Guide
!
Trainer Materials compact disc

To prepare for this module:
!
Read all of the materials for this module.
!

Complete the practices and lab.

Presentation:
150 minutes

Practices:
30 minutes

Lab:
60 minutes
Required materials
Preparation tasks
iv Module 5: Designing a Highly Available Physical Network


How to Teach This Module
This section contains information that will help you to teach this module.
Inform the students that each lesson in this module is a critical task in the
design process and that, at the end of the module, they will complete a lab that
ties all of the lessons (tasks) together. This knowledge will help them to stay
focused during instruction.
The instructional strategy for this module is to provide students with the
knowledge and skills that they need to design a highly available physical
network based on bandwidth requirements, redundant path requirements,
physical network topology requirements, and switch and router configuration
requirements of a highly available Web infrastructure.
Lesson: Selecting Networks Based on Bandwidth Requirements
This section describes the instructional methods for teaching each topic in this
lesson.
The overview page for this lesson introduces the need for selecting a physical

network based on the bandwidth requirements of a highly available Web
infrastructure. The instructional strategy for this lesson is to explain to students
that they can select the appropriate physical network by determining the
network throughput, the appropriate speed for the network, and the time needed
to download data from the network.
The topic pages for this lesson and the appropriate instructional strategy are
listed as follows:
The purpose of this page is to review the factors affecting network performance.
Tell students that network throughput plays an important role in the design of
any Web infrastructure and application, whereas bandwidth and latency affect
the choice of the client and application architecture.
This page introduces the two commonly used methods of determining the
network throughput: transactional cost analysis and default page analysis.
Explain to students that they need to know the number of bytes in a typical
client transaction and the maximum amount of concurrent transactions per
second going through the system to determine the amount of bandwidth needed
for a Web infrastructure.
The purpose of this page is to explain to students that they need to decide on the
appropriate network speed that is required to meet the business needs of the
Web infrastructure based on the required amount of network throughput and
available network speeds.
This page explains how to calculate the time required to download data from
the network to clients by using different download speeds. Emphasize to
students that they need to determine the maximum number of concurrent
connections and the amount of the data to be transmitted over the Web
infrastructure to calculate this value.
The guidelines page provides students with the action steps that they must
address before they can select a physical network based on the bandwidth
requirements of a Web infrastructure. Review the action steps with the students
and ensure that they understand how these steps map to the task. Also,

emphasize the importance of addressing all of these requirements.
Factors affecting
network performance
Methods for determining
the network throughput
Considerations for
determining the
appropriate speed for
the network
Time required to
download data from the
network
Guidelines for selecting
networks based on
bandwidth requirements
v


Practice: Select Networks Based on Bandwidth Requirements
You will divide the class into design teams. Give the students five minutes to
read through the scenario and the design considerations carefully before they
answer the questions. Tell the class that each team must be prepared to justify
their answers.
Lesson: Selecting Redundant Paths for Network Connectivity
This section describes the instructional methods for teaching each topic in this
lesson.
The overview page for this lesson introduces the need for selecting redundant
paths for network connectivity. The instructional strategy for this lesson is to
explain how to add redundancy in the design for the Internet connectivity and
the internal network connectivity to provide a greater level of fault tolerance.

The topic pages for this lesson and the appropriate instructional strategy are
listed as follows:
The purpose of this page is to explain that one way of making a physical
network highly available is to add redundant connections from a Web
infrastructure to the Internet. Tell students that when they design the external
network, they need to determine if a redundant Internet connection is possible
and what are the implications of using one Internet service provider (ISP) or
multiple ISPs.
This page explains that students need to focus on making the local area network
(LAN) more available for the Web infrastructure. Emphasize that the key to
making a LAN highly available is to configure backup paths to ensure there is
no single path from end-to-end in the infrastructure. Tell the students that they
can select either redundant Ethernet network paths or redundant Fiber
Distributed Data Interface (FDDI) network paths to make the internal network
highly available.
The guidelines page provides students with the action steps that they must
address before they can select redundant paths for network connectivity.
Review these steps with the students and ensure that they understand how these
steps map to the task. Emphasize the importance of addressing all of these
requirements.
Review: Selecting Redundant Paths for Network Connectivity
Give the students five minutes to read through the questions carefully before
they answer them. Tell students that they must be prepared to justify their
answers.
Considerations for
determining redundant
external network paths
Considerations for
determining redundant
internal network paths

Guidelines for selecting
redundant paths for
network connectivity
vi Module 5: Designing a Highly Available Physical Network


Lesson: Selecting the Appropriate Network Topology
This section describes the instructional methods for teaching each topic in this
lesson.
The overview page for this lesson introduces the need for selecting the
appropriate topology for the physical network. The instructional strategy for
this lesson is to explain to students that they can select the appropriate network
topology by selecting the appropriate Internet connection topology and the
appropriate topology for the internal network.
The topic pages for this lesson and the appropriate instructional strategy are
listed as follows:
This page introduces the types of ISPs that are available for hosting different
Web environments. Tell students that ISPs are broadly classified into in-house
Web hosting services and outsourced Web hosting services, and use the
PowerPoint slide for this page to differentiate between the characteristics of the
various ISP types.
The purpose of this page is to explain to students that they will need to select at
least two connections to the most reliable ISP, and another connection to a
separate ISP to ensure highly available Internet connectivity. Emphasize that
the Internet connectivity options for maintaining an in-house Internet data
center differ from those for an outsourced Internet data center.
This page explains to students that when selecting a high-speed network
topology for high availability, they can choose from four technologies: Gigabit
Ethernet, Fast Ethernet, FDDI, and asynchronous transfer mode (ATM).
Explain to students the considerations for selecting each of the four

technologies.
The guidelines page provides students with the action steps that they must
address before they can select the appropriate topology for the physical
network. Review the action steps with the students and ensure that they
understand how these steps map to the task. Also, emphasize the importance of
addressing all of these requirements.
Practice: Select the Appropriate Network Topology
You will divide the class into design teams. Give the students five minutes to
read through the scenario and the design considerations carefully before they
answer the questions. Tell the class that each team must be prepared to justify
their answers.
Types of Internet service
providers
Considerations for
determining the
appropriate Internet
connection topology
Considerations for
determining the
appropriate technology
for the internal network
topolo
gy

Guidelines for selecting
the appropriate network
topology
vii

Lesson: Selecting a Highly Available Switch Configuration
This section describes the instructional methods for teaching each topic in this
lesson.
The overview page for this lesson introduces the need for selecting a highly
available switch configuration for the physical network. The instructional
strategy for this lesson is to explain to students that they can provide high
reliability and fault tolerance in the network by determining the appropriate
switches and the appropriate switch configuration for the physical network.
The topic pages for this lesson and the appropriate instructional strategy are
listed as follows:
This page introduces the concept of broadcast and collision domains. Explain to
students that they must be cognizant of the broadcast and collision domains for
the network because these domains can impact the performance and the
perceived availability of the Web infrastructure.
The purpose of this page is to explain to students that they need to use
redundant or fault tolerant switches to provide high availability for the network
topology. Explain to students the importance of the factors affecting the
selection of appropriate switches for a physical network.
This page explains the characteristics of a highly available switch configuration.
Emphasize to students that when designing the layer 2 network for the Web
infrastructure, the scenario for the User Services and Business Logic tiers is not
the same as that in a conventional enterprise LAN; there are considerations,
such as the use of Network Load Balancing in each of these tiers, which will
influence their design. However, they will use the same design considerations
for the servers in the Data Services tier that they would use for an enterprise
LAN.
The purpose of this page is to explain to students that they need to use full-
duplex interfaces and Inter Switch Links and trunks. In addition, they must
ensure the availability of adequate switch buffer memory when selecting the
appropriate switch configuration for the physical network.

The guidelines page provides students with the action steps that they must
address before they can select the appropriate switch configuration. Review the
action steps with the students and ensure that they understand how these steps
map to the task. Emphasize to the students the importance of addressing all of
these requirements.
Review: Selecting a Highly Available Switch Configuration
Give the students five minutes to read through the questions carefully before
they answer them. Tell students that they must be prepared to justify their
answers.
Broadcast and collision
domains
Considerations for
selecting appropriate
switches
A highly available switch
confi
g
uration
Considerations for
selecting the
appropriate switch
confi
g
uration
Guidelines for selecting
the appropriate switch
confi
g
uration
viii Module 5: Designing a Highly Available Physical Network

Lesson: Selecting a Highly Available Routing Configuration
This section describes the instructional methods for teaching each topic in this
lesson.
The overview page for this lesson introduces the need for selecting a highly
available routing configuration for the physical network. The instructional
strategy for this lesson is to explain to students that when selecting the
appropriate routing configuration, it is recommended that they separate the
Business Logic and Data Services tiers from the User Services tier, but their
solution may require several subnets in each tier to achieve the required
performance and isolation.
The topic pages for this lesson and the appropriate instructional strategy are
listed as follows:
This page introduces the various routing protocols used in redundant IP routing
and dynamic routing protocols. Explain to students how each routing protocol
works.
The purpose of this page is to explain to students the characteristics of a highly
available IP routing configuration.
This page explains the considerations for determining the routing architecture
for the User Services tier. Explain to students the need for considering the
availability of the Internet connection from the Web infrastructure traffic, the
number of public addresses required for the Web infrastructure, and the use of
load-balanced multihomed configurations when selecting the routing
architecture for the User Services tier.
The purpose of this page is to explain to students the considerations for
determining the routing architecture for the Business Logic and Data Services
tiers. Emphasize to students the need for considering the types of hosts in the
Web infrastructure, the types of addresses required for the hosts, and the
number of subnets required for the Web solution when selecting the routing

architecture for the Business Logic and Data Services tiers.
The guidelines page provides students with the action steps that they must
address before they can select the appropriate routing configuration. Review the
action steps with the students and ensure that they understand how these steps
map to the task. Emphasize the importance of addressing all of these
requirements.
Review: Selecting a Highly Available Routing Configuration
Give the students five minutes to read through the questions carefully before
they answer them. Tell students that they must be prepared to justify their
answers.
Routing protocols for a
highly available physical
network
A highly available IP
routing configuration
Considerations for
determining the routing
architecture for the User
Services tier
Considerations for
determining the routing
architecture for the
Business Logic and
Data Services tiers
Guidelines for selecting
a highly available
routin
g
confi
g

uration
ix


Lab: Designing a Highly Available Physical Network
In this lab, students will design a highly available physical network to meet the
needs of the Government Portal scenario. Their design will include components
that meet the bandwidth requirements, redundant path requirements, physical
network topology requirements, and switch and router configurations
requirements of the given scenario. The students will then make appropriate
high availability recommendations for the design. As with the practices, you
will divide the class into design teams. Give the students 30 minutes to read
through the scenario and the design considerations carefully before they answer
the questions.
If white board space is available, have each team draw their design on the
board. Or, if Microsoft Visio
®
is available and the students are comfortable
using it, they could forward their design to you for display on the screen. Each
team must be prepared to justify their answers.
Depending on team experience, the Web infrastructure designs can be relatively
simple or quite complex. You may also discover that some features of their
Web infrastructure design may be incomplete or wrong because they do not
have the prerequisite knowledge. You should only focus on the part of the
design that addresses the lesson component being taught.
You can allow other teams to critique each design, but it is important that you
explain to the students that there are no wrong or right answers. What they need
to take from this exercise is the opportunity to practice their design ideas and
get peer review in a lab environment. Depending on business requirements,
their actual designs may vary.

Module 5: Designing a Highly Available Physical Network 1


Overview
!
Selecting Networks Based on
Bandwidth Requirements
!
Selecting Redundant Paths for
Network Connectivity
!
Selecting the Appropriate Network
Topology
!
Selecting a Highly Available Switch
Configuration
!
Selecting a Highly Available Routing
Configuration
Designing a Highly Available Physical
Network
Start End

*****************************
ILLEGAL FOR NON
-
TRAINER USE

******************************
The information in this module introduces the physical network of the Web
infrastructure and the factors that impact availability. You can design a highly
available physical network based on bandwidth requirements, redundant path
requirements, physical network topology requirements, and switch and router
configuration requirements of a highly available Web infrastructure.
After completing this module, you will be able to:
!
Select a network based on bandwidth requirements.
!
Select redundant paths for network connectivity.
!
Select appropriate networking topology for the physical network.
!
Select a highly available switch configuration for the physical network.
!
Select a highly available routing configuration for the physical network.
!
Select the appropriate technologies, configurations, and hardware
components required for designing a highly available physical network.

Introduction
Ob
j
ectives
2 Module 5: Designing a Highly Available Physical Network


Lesson: Selecting Networks Based on Bandwidth
Requirements

!
Factors Affecting Network Performance
!
Methods for Determining the Network Throughput
!
Considerations for Determining the Appropriate Speed
for the Network
!
Time Required to Download Data from the Network
!
Guidelines for Selecting Networks Based on
Bandwidth Requirements
Selecting Networks Based on Bandwidth Requirements

*****************************
ILLEGAL FOR NON
-
TRAINER USE
******************************
For a Web infrastructure to be highly available, you must ensure that there is
enough network bandwidth to satisfy the peak user load. If there is not enough
bandwidth to meet the needs of the users, packets queue up in routers and
switches and users experience either excessive wait times or server timeouts
when accessing the Web infrastructure.
Your physical layer network must be able to support your estimated peak
bandwidth requirements, and your design must have the desired redundancy
and reliable components to meet the level of availability that you want to
achieve.
This lesson will enable you to select a physical network based on the bandwidth
requirements of a highly available Web infrastructure.

After completing this lesson, you will be able to:
!
Identify the factors affecting network performance.
!
Describe the methods for determining the throughput for the physical
network.
!
Describe the considerations for determining the appropriate speed for the
physical network.
!
Calculate the time required to download data from the network based on the
throughput for the network and the network speed.
!
Select a network based on the bandwidth requirements of a Web
infrastructure.

Introduction
Lesson ob
j
ectives
Module 5: Designing a Highly Available Physical Network 3


Factors Affecting Network Performance
!
Network performance is affected by:
"
Bandwidth
"
Latency

"
Throughput

*****************************
ILLEGAL FOR NON
-
TRAINER USE
******************************
Networks present limitations to the amount of data that they can transfer in a
given period of time. When selecting a physical network for a highly available
Web infrastructure, you must be aware of the factors that affect the performance
of a network.
The primary factors that affect network performance are link speed
(bandwidth), delay (latency), and throughput.
!
Bandwidth is the maximum amount of data that can travel a
communications path in a given time, often measured in bits per second.
!
Latency is the amount of time that it takes a packet to travel from source to
destination, usually measured in seconds.
!
Throughput is the data transfer rate that is achieved by combining the
effects of bandwidth and latency. Simply put, bandwidth is what you pay
for, and throughput is what you actually receive.

Although the two key network characteristics, bandwidth and latency, affect
your choice of the client and application architecture, network throughput plays
an important role in the design of any Web infrastructure and application. The
throughput, or capacity of a pipe, is often shown numerically as bandwidth
multiplied by the round-trip time (latency). This value is known as the

bandwidth-delay product. The Transmission Control Protocol (TCP) window
size is automatically adjusted to suit the bandwidth-delay product to achieve
best performance over a particular link, but Internet-based clients usually
negotiate very small window sizes because of the relatively small transfers of
data to the client.
Introduction
Factors that affect
network performance
4 Module 5: Designing a Highly Available Physical Network


Clients working across local area networks (LANs) in an intranet typically have
a high level of network throughput available. Multi-megabits per second
(Mbps) bandwidths are available with very small latency, often less than 25
milliseconds round trip. Internet-based clients, however, can expect lower
levels of throughput to be available with a wider variance in network quality of
service because of the multiple hops and greater distances involved. While
some clients are connected across dial-up telephone lines, others could be
connected over cable modems, digital subscriber lines (DSL), or better services.
Highly interactive applications place greater demands on both network
bandwidth and latency. To deliver adequate performance with low available
bandwidth, you may have to design your applications with a less interactive
interface. It may be necessary to move portions of the User Services
responsibilities to the client. This move, in addition to some form of caching of
the query or client data, could yield acceptable response times. Client-side
controls and applications that can provide presentation responsibilities may be
better adapted to work in low-bandwidth situations.
For example, consider a browser-based application that displays a hierarchy of
information as a tree in the client, which shows a navigational menu or a list of
mail folders. If the network is fast, it may be acceptable to make a request to the

server every time the user selects a different node to see a list of the node’s
child elements. The server then dynamically generates a new screen reflecting
the expanded node. However, if the network throughput is low because of high
latency or restricted bandwidth, a better approach might be to have the client
cache the entire node hierarchy locally.
Module 5: Designing a Highly Available Physical Network 5


Methods for Determining the Network Throughput
!
Transactional cost analysis
"
Determine the throughput for the network by multiplying
the amount of data to be transmitted by the amount of
concurrent connections expected during peak usage
!
Default page analysis
"
Determine the throughput for the network by multiplying
the size of the default Web page by the number of
concurrent users per second

*****************************
ILLEGAL FOR NON
-
TRAINER USE
******************************
When selecting a network for a highly available Web infrastructure, you must
know the maximum amount of data during peak load that users need to access
in each tier of the infrastructure. By knowing this value, you can correctly select

the network speed that your system requires to maintain acceptable network
throughput on each tier in your Web infrastructure.
You can determine the amount of throughput needed for the Web infrastructure
in one of two ways: transactional cost analysis and default page analysis. You
must know the number of bytes in a typical client transaction and the maximum
amount of concurrent transactions per second going through the system. A
typical client transaction can include a client sending their credit card
information to the server, or another Internet server supplying service
information to your Web site.
One of the methods to determine the network throughput is to use a
transactional cost analysis approach. To use this approach, you must determine
a typical client transaction, such as browsing an online catalog or making a
purchase from a shopping cart, which you will use for analysis. You must also
determine what a typical client will do on your Web site, and then capture the
amount of bandwidth that the system will use to perform the transaction. You
must also analyze the page mix that is requested because different types of
pages, such as catalog browsing, shopping cart updates, user account updates,
and purchasing, have different bandwidth needs.
Introduction
Transactional cost
analysis
6 Module 5: Designing a Highly Available Physical Network


When a higher percentage of visitors complete orders, the page mix being
requested has a larger component of pages with higher transactional cost or
higher volumes of network traffic. Catalog browsing pages can use more traffic
on the User Services network but less traffic on the Data Services network
because the pages are typically cached on the Web servers, thereby reducing
database access. However, user account pages can use more bandwidth because

they require database access and use Secure Sockets Layer (SSL), which can
use much more bandwidth on the User Services and the Data Services network,
thereby making this type of transaction more expensive to process than
unsecured pages. Finally, checkout pages can be the most expensive because
they cause more expensive database transactions and Business Logic workflow
with credit card validation than user account pages.
Typically, all Web-based electronic commerce sites service customer requests
that fall into one of the categories in the following table.
Type of usage Percentage of traffic

Browse 80%
Search 9%
User registration 2%
Add item 5%
Buy 4%

After you determine the amount of data that the system will transmit for a given
scenario, multiply that amount by the number of expected concurrent
connections to calculate the required system throughput in each network tier.
This approach will give you the best result for each network in an n-tier
architecture; however, it may be difficult to determine the scenario that most
users commonly use.
Another simple method for determining the amount of bandwidth needed for a
Web infrastructure is to calculate the size of the default Web page and multiply
it by the number of concurrent users. This method works well for determining
the network usage on the User Services network if the size of the default Web
page is an average size page for the entire Web site. However, default page
analysis will not give you accurate information about bandwidth usage of the
Data Services network.
After you decide which approach to use for determining the network

throughput, you can use a performance monitor or network monitor to
determine the amount of traffic traveling across the network during your test of
a single client transaction. When you determine the amount of bandwidth used
for a single transaction, you can then extrapolate the single transaction to the
amount of concurrent connections during peak usage. You can use this number
to calculate the throughput for the network, which will help you to determine
the network speed required for meeting business needs.
Default pa
g
e anal
y
sis
Module 5: Designing a Highly Available Physical Network 7


Considerations for Determining the Appropriate Speed for the
Network
When selecting the appropriate network connection type:
!
Determine the network bandwidth based on the
estimated future growth
!
Determine the network speed required for the Web
infrastructure
1

G
b
p
s

1
0

M
b
p
s


1
0
0

M
b
p
s


<

5
0
%

U
t

i
l
i
z
a
t
i
o
n

*****************************
ILLEGAL FOR NON
-
TRAINER USE
******************************
After determining the required amount of network throughput, you must decide
on the network speed that is required to meet your business needs. To select the
network speed, use the information gathered from your network throughput
calculations, allow for growth, and leave sufficient bandwidth for unexpected
network traffic.
After you have determined the throughput required for your Web infrastructure,
you must determine how much network bandwidth you want to be made
available for future growth. For example, if you want 60 percent of the
available bandwidth of your network and you forecast future growth to be
50 percent of a 10-Mbps line, you will need to upgrade your network to the next
available speed of 100 Mbps speed.
When you calculate growth, you must be aware that an increase of 50 percent
on the User Services tier might not result in an increase of 50 percent on the
Data Services tier. Each tier in an n-tier environment has its own network
characteristics and might yield different results when calculating future growth.

In fact, a 50 percent increase in the User Services tier may only yield a
25 percent increase in the Data Services tier, because many images and data get
cached on the Web servers.
Introduction
Network bandwidth
8 Module 5: Designing a Highly Available Physical Network


After allowing for enough bandwidth and growth, you can select the
appropriate LAN and wide area network (WAN) speeds for the Web
infrastructure. Network speeds vary for LANs and WANs.
The most common network speeds available for LANs range from 10 Mbps to
1 gigabit per second (Gbps), as listed in the following table.
LAN connection type Speed

Standard Ethernet 10 Mbps
Fast Ethernet 100 Mbps
Fiber Distributed Data Interface (FDDI) 100 Mbps
Gigabit Ethernet 1,000 Mbps or 1 Gbps

The most common WAN transmission speeds available range from 1.54 Mbps
for T1 to 2488.5 Mbps for optical carrier 48 (OC48).
WAN connection type Speed

T1 or Digital Services-1 (DS-1) 1.54 Mbps
T3 or DS-3 44.74 Mbps
OC1 51.84 Mbps
OC3 155.53 Mbps
OC12 622.13 Mbps
OC48 2488.5 Mbps

Network speeds
Module 5: Designing a Highly Available Physical Network 9


Time Required to Download Data from the Network
Download time in seconds
Concurrent connections at 28.8 Kbps
60—
55—
50—
45—
40—
35—
30—
25—
20—
15—
10—
5—
10 20 30 40 50 60 70 80 90 100 1550 1560 2000
25 sec minimum 28.8 Kbps
T1 (client-side speed) T1 saturates at
54 connections
47.75 sec at
100 connections
T3 saturates at
1550 connections
33 sec at
2000 connections

Page size
720 Kilobits

*****************************
ILLEGAL FOR NON
-
TRAINER USE
******************************
Many Web sites are designed to respond to the user with a reasonable download
speed. The WAN and LAN traffic of the Web infrastructure can affect the end
user experience if the LAN and WAN get congested to the point where the
client cannot obtain the data at the end user’s connection speed. For example, if
the user has a 28.8 kilobits per second (Kbps) modem and is only receiving data
at 15.6 Kbps because the WAN link to the Web site is congested, the end user
will think that the site is neither highly available nor very reliable.
To calculate the time required to download data from your network to the
client, you must determine the maximum number of concurrent connections and
the amount of the data to be transmitted over the Web infrastructure. Consider a
Web page that is 90 kilobytes (KB), or 720 kilobits, in size, including overhead.
This page takes 25 seconds to download through a 28.8-Kbps connection if
everything is working perfectly and you ignore application latencies, which
typically add a few seconds. However, the page will take longer to download if
there is any bottleneck at the server, the network is overloaded or slow, or the
user’s connection is slower than the full 28.8 Kbps because of poor line quality.
It will take about 54 connections at 28.8 Kbps to saturate the capacity of a T1
line with a WAN speed of 1.54 Mbps. If no more than 54 clients simultaneously
request the hypothetical 90-KB page, and if the server can perform the
processing required to keep up with the requests, all of the clients will receive
the page in the 25 seconds calculated in the example, ignoring application
latencies. However, if 100 clients simultaneously request the same page, the

amount of data to be transferred is 100 times 720 kilobits, which takes between
47 and 48 seconds to travel down a T1 line.
Introduction
Determining the time
required to download
data
10 Module 5: Designing a Highly Available Physical Network


A T3 line carries nearly 45 Mbps, about 30 times as much capacity as a T1 line,
and it takes almost 1,550 clients at 28.8 Kbps to saturate its bandwidth. At
2,000 simultaneous connections, it still takes less than 33 seconds to download
the page. It is recommended that the time required to transmit the largest page
between your User Services network and your Internet service provider (ISP)
does not exceed 3 seconds during average load and is not be greater than 5
seconds during peak load.
The following table displays the estimated time for downloading different file
sizes over various client connection types.

File size
28.8 Kbps
modem
56 Kbps
modem
512 Kbps
cable modem

T1 line

30 KB 8 sec 4 sec <1 sec <1 sec

500 KB 2 min 22 sec 1 min 13 sec 8 sec 2 sec
1 MB 4 min 51 sec 2 min 29 sec 16 sec 5 sec
5 MB 24 min 16 sec 12 min 28 sec 1 min 21 sec 26 sec


Module 5: Designing a Highly Available Physical Network 11


Guidelines for Selecting Networks Based on Bandwidth
Requirements
!
Determine the throughput for the network by:
"
Determining the number of clients to support
"
Defining a standard client
"
Defining a standard transaction
!
Determine the appropriate speed for the network
by determining the:
"
Projected bandwidth required for future growth
"
Available network speeds
!
Determine the time required to download data
from the network

*****************************

ILLEGAL FOR NON
-
TRAINER USE
******************************
When designing a highly available physical network for a Web infrastructure,
you must first need to select a network based on the bandwidth requirements of
the Web infrastructure.
As you select internal networks based on your bandwidth requirements, apply
the following guidelines:
!
Determine the throughput for the network. To determine the throughput
required for the network, you need to:
• Determine the number of clients the Web infrastructure can support.
• Define a standard client. By defining a standard client, you can
determine the throughput for the typical client access.
• Define a standard transaction. By defining a standard transaction, you
can determine the throughput that client access causes to the Business
Logic and Data Services tiers.
!
Determine the appropriate speed for the network. To determine the
appropriate network speed required to meet the throughput for the network,
you need to determine the:
• Projected bandwidth necessary to accommodate future network growth.
• Available network speeds.
!
Determine the time required to download data from the network. You must
determine the time required to download data from the network based on the
network throughput and the available network speeds. You must ensure that
the time required by the client to download data from the Web infrastructure
does not exceed your maximum permissible value.

Introduction
Design guidelines
12 Module 5: Designing a Highly Available Physical Network


Practice: Select Networks Based on Bandwidth Requirements
In this practice, you will:
#
Select a physical network based on the
bandwidth requirements of a given business
scenario

*****************************
ILLEGAL FOR NON
-
TRAINER USE
******************************
You are the Web infrastructure administrator for Contoso, Ltd. You need to
calculate latency and bandwidth requirements for the company Web site. The
Web site is expected to have the following characteristics:
!
The average connection speed of a customer visiting the Web site is
expected to be 28.8 Kbps.
!
The expected peak concurrent user load is 200.
!
The user load is expected to double in the next two years.
!
Dual, redundant connections to the Internet are required. This requirement

could be provided through a single ISP with redundant connections to the
Internet.
!
A single Internet connection must be able to handle peak user traffic during
the next two years.

The expected size and the download frequency of various types of Web pages
are shown in the following table.
Type of Web page Average size Download frequency

Home page 50 KB 20%
Specials page 100 KB 10%
Product catalog page 40 KB 60%
Detailed product images 200 KB 10%

Scenario
Module 5: Designing a Highly Available Physical Network 13


1. Under optimum network conditions, what is the latency for the average
customer for each page type?
a. Home page
Approximately 14 seconds. (Convert the average size of the Web
page to kilobytes by multiplying by 8 and then dividing by the
average connection speed of 28.8 Kbps.)


b. Specials page
Approximately 28 seconds.

c. Product catalog page
Approximately 11 seconds.


d. Detailed product image
Approximately 56 seconds.


2. What are the bandwidth requirements for the Internet connections?
The average page size is 64 KB, which is derived as follows:
(50 KB x 0.2) + (100 KB x 0.1) + (40 KB x 0.6) + (200 KB x 0.1) = 64 KB
Average concurrent user load is approximated as follows:
64 Kb (av. page) * 8 (bits/byte) * 200 (concurrent users) = 102.4 Mbps
This bandwidth requirement could be met by a single optical carrier 3
(OC3) (155 Mbps) connection or by multiple T3 (43.232 Mbps)
connections.
By adding 100 percent growth over the next two years, the required
bandwidth for a single ISP connection grows to 205 Mbps. To meet this
bandwidth requirement, you could use two OC3 connections to achieve
the Internet connection redundancy requirement. A single ISP may be
able to provide a service level agreement (SLA) that guarantees the
required bandwidth, thus eliminating the need to provide redundant
connections from the site.





Questions

14 Module 5: Designing a Highly Available Physical Network


Lesson: Selecting Redundant Paths for Network
Connectivity
!
Considerations for Determining Redundant External
Network Paths
!
Considerations for Determining Redundant Internal
Network Paths
!
Guidelines for Selecting Redundant Paths for Network
Connectivity
Selecting Redundant Paths for Network Connectivity

*****************************
ILLEGAL FOR NON
-
TRAINER USE
******************************
When designing a highly available physical network, you must determine if
redundant paths are required for network connectivity. Using redundant paths
for network connectivity is an absolute requirement for delivering high
availability in a Web infrastructure. By using redundant network paths, you can
negotiate and route around bottlenecks and temporary failures.
This lesson will discuss how to add redundancy in your design for both the
Internet connectivity and the internal network connectivity to provide a greater
level of fault tolerance.
After completing this lesson, you will be able to:

!
Describe the considerations for determining if single or multiple paths are
required or appropriate for Internet connectivity.
!
Describe the considerations for determining if single or multiple paths are
required or appropriate for internal network connectivity.
!
Select redundant paths for Internet and internal network connectivity.


Introduction
Lesson objectives
Module 5: Designing a Highly Available Physical Network 15


Considerations for Determining Redundant External Network Paths
You can ensure redundant Internet connection by:
!
Having more than one network connection to the
Internet
!
Selecting more than one Internet service provider
Switch
Switch
Switch
Switch
Router
Router
Router
Router

ISP
ISP

*****************************
ILLEGAL FOR NON
-
TRAINER USE
******************************
After you have determined the amount of bandwidth that your system needs to
support your Web infrastructure during peak load, you can then focus on how to
make the physical network highly available. One way to make a physical
network highly available is by adding redundant connections from your Web
infrastructure to the Internet. When designing the external network, you will
need to determine if a redundant Internet connection is possible and what are
the implications of using one ISP or multiple ISPs.
Whether your Web hosting infrastructure is in your data center or your servers
are at an off-site location that an Internet data center hosts, you must have
network connectivity to the Internet. If the connectivity to the Internet is down,
your Web site is obviously inaccessible. Your connection to the Internet must
be highly available.
You can ensure that your Internet connection is highly available by:
!
Having more than one network connection to the Internet.
• Make sure that there is sufficient bandwidth on each Internet connection
so that if one of the connections fails, the others can handle the peak user
load.
• Make sure that each router has a secondary route for redirecting traffic in
case of a line failure. If the link to the Internet fails, the router can
redirect outgoing traffic to the existing router until the connection to the
Internet is restored.

!
Selecting more than one ISP.
• It may not be enough to have redundant Internet connections if your ISP
goes offline. It is recommended that you use two separate ISPs for your
Internet connectivity.
• If you have two ISPs, you must have separate networks. However, if you
have only one subnet, you must use network address translation (NAT)
to hide the actual Internet Protocol (IP) addresses that the original ISP
provided to you.
Introduction
Redundant Internet
connection