Tài liệu Module 2: Designing Highly Available Web Infrastructure Servers doc

Contents
Overview 1
Lesson: Choosing a Hardware Platform 2
Lesson: Determining the Windows 2000
Operating System Configuration 24
Lesson: Selecting the Appropriate
Technologies Based on Microsoft Internet
Information Services 5.0 43
Lesson: Designing a Highly Available
COM+ Environment Using Microsoft
Application Center 2000 55
Lesson: Designing a Highly Available
Firewall That Includes Server Publishing 71
Lesson: Designing a Highly Available
Database Solution Using Microsoft
SQL Server 2000 86
Lab: Designing Highly Available Web
Infrastructure Servers 101

Module 2: Designing
Highly Available Web
Infrastructure Servers


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, place or event 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 United States and/or other countries.

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


Module 2: Designing Highly Available Web Infrastructure Servers iii


Instructor Notes
This module provides students with the knowledge and skills that they will need
to design a complex installation of multiple products on a single server. After
completing this module, students will be able to:

!
Choose a hardware platform.
!
Determine the Microsoft
®
Windows
®
2000 operating system configuration
for the Web infrastructure servers.
!
Select the appropriate technology for a highly available solution based on
Microsoft Internet Information Services (IIS).
!
Design a highly available COM+ (Component Object Model) Environment
by using Microsoft Application Center 2000.
!
Design a highly available firewall solution by using Microsoft Internet
Security and Acceleration (ISA) Server.
!
Design a highly available database solution by using Microsoft
SQL Server
™
2000.

To teach this module, you need the following materials:
!
Microsoft PowerPoint
®
file 2088A_02.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:
270 minutes

Practices:
90 minutes

Lab:
60 minutes
Required materials
Preparation tasks
iv Module 2: Designing Highly Available Web Infrastructure Servers


How to Teach This Module
Ensure that the student understands that each lesson is in this module is a
critical task in the design process and at the end of the module they will
complete a lab that helps to tie all of the lessons (tasks) together. This will help
the student to stay focused during instruction.
The instructional strategy for this module is to provide the students with the
knowledge and skills needed to design a highly available Web infrastructure by

using Microsoft products and technologies.
Lesson: Choosing a Hardware Platform
This lesson teaches the students how to choose the appropriate hardware
platform for a highly available Web infrastructure. They will need to ensure that
the computer hardware platform is highly available by meeting the reliability
needs for their Web design.
The students must determine the level of availability required for each server in
the infrastructure. This is a high level overview of required availability,
including the following topics:
!
Planned and unplanned system downtime
!
Calculated values
!
Document and specify system availability

Students will learn to choose system components that make a physical server
highly available. The most important feature of this page is the table that
identifies the various system components and the considerations the students
will need to have before they can design a highly available system. Use the
table as a review. You need to be familiar with:
• Choosing a hardware configuration

Students will learn to choose highly reliable computer components. The most
important feature of this page is the table that identifies the various computer
components and the considerations the students need before they can make
design decisions that ensure avoidance of single points of failure. Use the table
as a review. You need to be familiar with:
• Reliability of the hardware components

Students will select the appropriate network configuration to improve data
availability. Review the best practices for designs that include multiple network
adapters. You need to be familiar with:
• Teaming network adapters

Required level of
availability
Highly available system
components
Highly available
computer components
Network configuration
Module 2: Designing Highly Available Web Infrastructure Servers v


To ensure that the computer hardware platform is highly available, the
computer hardware must be highly reliable. Use the slide to review some of
more common computer hardware issues. You need to be familiar with:
!
Server architecture
!
Fans
!
Peripheral Component Interconnect (PCI) hot plug
!
Power supply needs
!
Random access memory

This page provides an overview of disk configurations in a highly available

Web infrastructure. Use it as a review. You need to be familiar with:
!
Operating system disks
!
Disk configurations for transactional applications

For students who are not familiar with Microsoft server cluster, this page
provides an overview of the technology. Use it as a review. You need to be
familiar with:
!
Disabling network basic input/output system (NetBIOS) over Transmission
Control Protocol/Internet Protocol (TCP/IP)
!
Domainlets
!
Hardware Compatibility List (HCL) hardware
!
Private and mixed networks
!
Quorum disk

You can use the best practices on this page to ensure that the student
understands how they can choose the appropriate hardware for a server cluster.
You need to be familiar with:
• Server hardware for Microsoft Cluster service

For students who are not familiar with Network Load Balancing, this page
provides an overview of the product. Review the best practices with the
students. You need to be familiar with:
!

Balancing Internet traffic
!
Cluster subnets

You can use the best practices on this page to ensure that the students
understand how to choose the appropriate hardware for a server cluster. You
need to be familiar with:
• Server hardware for Network Load Balancing

The guidelines page provides the students with the action steps they must
address before they can choose a computer hardware platform for a highly
available Web infrastructure. You should review the action steps with the
students and ensure that they understand how these steps map to the task. You
should emphasize to the students the importance of addressing all of these
requirements.
Highly available
computer hardware
Disk confi
g
uration
Microsoft server cluster
Server hardware for
server clusters
Microsoft Network Load
Balancin
g

Server hardware for
Network Load Balancing
Guidelines for choosing

a computer hardware
platform
vi Module 2: Designing Highly Available Web Infrastructure Servers


Practice: Choose a Hardware Platform
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: Determining the Windows 2000 Operating System
Configuration
This lesson teaches the students how to determine the appropriate
Windows 2000 operating system for the Web infrastructure servers. They will
learn the product features, processes for installing the operating system, and
how to determine the domain roles. You should be familiar with all of the
Microsoft products and technologies used in this module.
This page provides a high level overview of all of the Microsoft products that
are discussed in this course. Use the table provided to review the features of
these products that support a highly available Web infrastructure. You need to
be familiar with:
!
Features of Windows 2000
!
Microsoft Windows 2000 Advanced Server
!
Microsoft Windows 2000 Datacenter Server

Emphasize to students the importance of ensuring that the disk configuration for
their design is highly available before they can select the appropriate operating

system. You need to be familiar with:
!
Spindles
!
Logical disk volumes
!
NTFS file system (NTFS)

This page covers all of the recovery options for the Windows 2000 operating
system. Use the slide to provide a summary of the components. You need to be
familiar with:
!
Recovery console
!
Symbol files
!
Symbol server
!
Dump-options

Product features
Highly available disk
confi
g
uration
Windows 2000 operating
system components
Module 2: Designing Highly Available Web Infrastructure Servers vii

When using the Windows 2000 operating system, the students will need to
determine if the servers will be in a workgroup or a domain. The students
should know most of this content. Emphasize the concept of Group Policy
Loopback processing because the students may not be as familiar with this
topic. In addition, you should be familiar with:
!
Workgroup and domain membership
!
Account and server domains
!
Server types
!
Organizational units
!
Group policy and Group Policy object (GPO) usage
!
Group Policy Loopback processing

This page reviews two processes for installing the OS, script and image
installation. Use the slide to review the page content. You need to be familiar
with:
!
Script installation
!
Image installation

This page provides students with some best practices to follow when
configuring a highly available network. Use the slide to review the relevant
content on the page. You need to be familiar with:
!

Using full-duplex connections
!
Selecting path maximum transmission unit

This page reviews the network security features of Windows 2000. Use the
slide to review the relevant page content. Be prepared to answer any questions
the student might have about the following topics:
!
Filtering
!
Internet Protocol security (IPSec)
!
Service bindings

The guidelines page provides the students with the action steps they must
address before they can determine the operating system configuration for a
highly available Web infrastructure. You should review the action steps with
the students to ensure that they understand how these steps map to the task. You
should emphasize to the students the importance of addressing all of these
requirements.
Practice: Determine the Windows 2000 Operating System
Configuration
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.
Domain roles
Processes for installing
the operatin
g

s
y
stem
A highly available
network confi
g
uration
Network security
features
Guidelines for
determining the
operating system
configuration
viii Module 2: Designing Highly Available Web Infrastructure Servers


Lesson: Selecting the Appropriate Technologies Based on
Microsoft Internet Information Services (IIS) 5.0
This lesson teaches the students how to select the appropriate technologies
based on Internet Information Services 5.0 (IIS). They will need to identify
specific characteristics of the applications before they can determine the
technologies that will support their highly available Web infrastructure design.
Use the table on this page to provide an overview of the application
characteristics of Web applications and how they relate to IIS. The students
should be familiar with the application characteristics.
The only feature on this page that the students may not be familiar with is the
Application Center 2000 HTTP Request Forwarder. Be prepared to discuss this
concept. In addition, you need to be familiar with:
!
Network Load Balancing

!
Server cluster
!
Application Center 2000 HTTP Request Forwarder
!
Scaling

Use the table on this page to provide an overview of our recommended
solutions when using Microsoft products and technologies to support Web
applications. You need to be familiar with:
• Recommended solutions

The guidelines page provides the students with the action steps they must
address before they can select the technologies for a highly available IIS
solution. You should review the action steps with the students and ensure that
they understand how these steps map to the task. You should emphasize to the
students the importance of addressing all of these requirements.
Practice: Select the Appropriate Technologies for a Highly Available
IIS Solution
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.
Characteristics of the
application
Microsoft products and
technolo
g
ies
Microsoft products and

technologies to support
Web applications
Guidelines for selecting
the appropriate
technologies for a highly
available IIS solution
Module 2: Designing Highly Available Web Infrastructure Servers ix


Lesson: Designing a Highly Available COM+ Environment Using
Microsoft Application Center 2000
This lesson teaches the students how to design a highly available COM+
(Component Object Model) environment by using Application Center 2000.
They will learn how to manage content and configuration settings on a single
server, which they then replicate to other computers in a cluster.
Emphasize to students the concepts on this page by using the product features
bullet points and the hardware components, in addition to the recommendations
table. You need to be familiar with:
!
Example of Application Center 2000 deployment
!
Product features
!
Hardware requirements

Use the slide to review the major points on the page. You need to be familiar
with the following concepts:
!
Shared nothing architecture
!

Automated control of servers

Emphasize to students why using component load balancing to balance
component requests across a COM+ component application cluster provides
fault tolerance. Use the slide to facilitate student comprehension. You need to
be familiar with:
!
COM+ and component load balancing service
!
Automated control of servers

Be able to explain why implementing business logic as COM+ components
allows significant improvements in security, scalability, and performance. You
need to be familiar with:
!
Enabling component load balancing
!
Component load balancing architecture

You must be familiar with the three primary component load balancing models.
Discuss with students the reasons for separating and not separating the Web and
COM+ tiers. You need to be familiar with:
!
Component load balancing models
!
Wed and COM+ server tiers

Use the slide to demonstrate how Application Center 2000 supports high
availability. You need to be familiar with data collectors.

The guidelines page provides the students with the action steps they must
address before they can design a highly available COM+ environment. You
should review the action steps with the students and ensure that they understand
how these steps map to the task. You should emphasize to the students the
importance of addressing all of these requirements.
Application Center 2000
deployment
Application
requirements
COM+ routing
COM+ component
requirements
COM+ load balancing
Application Center 2000
supports high
availability
Guidelines for designing
a highly available COM+
environment using
Application Center
x Module 2: Designing Highly Available Web Infrastructure Servers


Practice: Design a Highly Available COM+ Environment Using
Microsoft Application Center 2000
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: Designing a Highly Available Firewall that Includes Server

Publishing
This lesson teaches the students how to design a highly available firewall. To
minimize the potential for attacks against the network, the design should
include a computer (router or firewall device) running a firewall process that
passes or translates traffic between the protected and unprotected networks
based on a set of defined rules and mechanisms.
The students should be familiar with the topics on this page. You can use the
slide as a review. You need to be familiar with the following topics:
!
Advanced mechanisms
!
Packet filters
!
Circuit proxy
!
Application gateways

This page introduces ISA Server and emphasizes the features and functionality
of the product. You can use the slide to cover the major topics on the page. You
need to be familiar with the following concepts:
!
ISA Server firewall features
!
ISA Server publishing
!
Scaling ISA Server
!
ISA Server and Active Directory
™
directory service

!
ISA Server load balancing

Use the graphic to demonstrate the recommended two-firewall solution. This
firewall architecture was chosen to provide isolation and security for the Web
infrastructure. Review the list of advantages and disadvantages with the
students. Emphasize how the advantages of this solution outweigh the
disadvantages. You need to be familiar with:
!
Single points of failure
!
Automated control of servers

Firewalls
Microsoft Internet
Security and
Acceleration Server
2000
Recommended firewall
solution
Module 2: Designing Highly Available Web Infrastructure Servers xi


The students must be able to determine the required number of public and
private addresses for the hosts. Emphasize the need for redundant security
devices and paths, and for reducing the required number of addresses. You need
to be familiar with:
!
Security devices and paths
!

Number of addresses

Because n-tier is the architecture for the Web infrastructure design, this page
discusses isolating the User Services tier (front-end). Use the slide to review the
content on the page, and use the design process bullets to emphasize the
importance of using a process to complete the design. You need to be familiar
with the following concepts:
!
Document data
!
Design process

The student will learn how to isolate the back-end (Business Logic and Data
Services) tiers. Use the slide to review the page content and emphasize as with
the previous topic the design process. You need to be familiar with:
!
Document required data
!
Process to complete design

The guidelines page provides the students with the action steps they must
address before they can design a highly available firewall solution. You should
review the action steps with the students and ensure that they understand how
these steps map to the task. You should emphasize to the students the
importance of addressing all of these requirements.
Practice: Design a Highly Available Firewall Solution that Includes
Server Publishing
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.
Required number of
public and private
addresses
User Services tier
Business Logic and
Data Services tiers
Guidelines for designing
a highly available
firewall solution that
includes server
publishing
xii Module 2: Designing Highly Available Web Infrastructure Servers


Lesson: Designing a Highly Available Database Solution Using
Microsoft SQL Server 2000
This lesson teaches the students how to design a highly available database
solution by using SQL Server 2000. They will learn how to select a network
strategy and the appropriate disk architecture for a SQL Server solution.
Use the slide to review the features and functionalities of SQL Server 2000.
You need to be familiar with database technologies.
You can use the slide to review the major topics on the page. Emphasize that
the recommended configuration is SQL Server 2000 with failover clustering.
You need to be familiar with the following topics:
!
Recommended configuration
!
Cluster service
!

Failover
!
Network Load Balancing

Although you should discuss all of the topics listed, focus on the best practices
listed on the slide. You need to be familiar with:
!
Communications configuration
!
Network protocols
!
Trusted connections to SQL Server
!
Client connection accounts
!
Separate query network

Use the slide to review the major points in the text. You need to be familiar
with the following topics:
!
Network strategies
!
Data security

Use the slide to review the text. Be prepared to answer questions about the
benefits of using redundant array of independent disks (RAID), and be familiar
with:
!
Disk architecture
!

Benefits of using RAID
!
Hardware system
!
System area networks
!
Storage area networks

SQL Server 2000
database technologies
SQL Server 2000
architecture
Communications
configuration
Network strategy
Disk architecture
Module 2: Designing Highly Available Web Infrastructure Servers xiii


The guidelines page provides the students with the action steps they must
address before they can design a highly available database solution by using
SQL Server 2000. You should review the action steps with the students and
ensure that they understand how these steps map to the task. You should
emphasize to the students the importance of addressing all of these
requirements.
Practice: Design a Highly Available Database Solution Using
SQL Server 2000
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.
Lab: Designing Highly Available Web Infrastructure Servers
In this lab, the student will design a highly available Web infrastructure server
by using Microsoft products and technologies to meet the needs of the
Government Portal scenario. The students will make appropriate high
availability recommendations for the design where required. 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, each team should be required to put their
design on the board. If Microsoft Visio
®
is available and the students are
comfortable using it, you could have them send their design to you for display
on the screen. Each team should be prepared to justify their answers.
Depending on team experience, the Web designs can be relatively simple or
quite complex. You may also discover that some features of their Web 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 the other teams to critique each design, but it is important that
you explain to the students that are no wrong or right answers. What they need
to take from this exercise is the opportunity to practice their design ideas and
obtain peer review in a lab environment. Depending on business requirements,
their actual design may vary.
Guidelines for designing
a highly available
database solution using
SQL Server 2000

Module 2: Designing Highly Available Web Infrastructure Servers 1


Overview
Start End
Designing Highly Available Web
Infrastructure Servers
!
Choosing a Hardware Platform
!
Determining the Windows 2000 Operating
System Configuration
!
Selecting the Appropriate Technologies
Based on Microsoft Internet Information
Services 5.0
!
Designing a Highly Available COM+
Environment Using Microsoft Application
Center 2000
!
Designing a Highly Available Firewall That
Includes Server Publishing
!
Designing a Highly Available Database
Solution Using Microsoft SQL Server 2000

*****************************
ILLEGAL FOR NON
-

TRAINER USE
******************************
The servers in your solution can be required to support several Microsoft
products to deliver the required services. After completing this module, you
will be able to apply the considerations for using Microsoft products and
technologies as components of a highly available Web infrastructure design.
Microsoft
®
Windows
®
2000 Server, Microsoft Advanced Server, and Microsoft
Windows 2000 Datacenter Server are the foundation for all server functions.
Additional products, such as Microsoft Application Center 2000 and Microsoft
SQL Server
™
2000 provide increased functionality.
At the end of each lesson, you will be provided with the considerations and
guidelines for the product highlighted in that lesson.
After completing this module, you will be able to:
!
Choose a hardware platform.
!
Determine the operating system configuration for a highly available Web
infrastructure.
!
Select the appropriate technologies for a highly available solution based on
Microsoft Internet Information Services (IIS).
!
Design a highly available COM+ (Component Object Model) environment
by using Application Center 2000.

!
Design a highly available firewall solution by using Microsoft Internet and
Security Acceleration (ISA) Server.
!
Design a highly available database solution by using SQL Server 2000.
!
Select the appropriate Microsoft products and technologies for designing
highly available Web infrastructure servers.

Introduction
Objectives
2 Module 2: Designing Highly Available Web Infrastructure Servers


Lesson: Choosing a Hardware Platform
!
Required Level of Availability
!
Highly Available System Components
!
Highly Available Computer Components
!
Network Configuration
!
Highly Available Computer Hardware
!
Disk Configuration
!
Microsoft Server Cluster
!

Server Hardware for Server Clusters
!
Microsoft Network Load Balancing
!
Server Hardware for Network Load Balancing
!
Guidelines for Choosing a Computer Hardware Platform
Choosing a Hardware Platform

*****************************
ILLEGAL FOR NON
-
TRAINER USE
******************************
This lesson will enable you to choose the appropriate hardware platform for the
servers in a highly available Web infrastructure. The hardware platform and the
components that you use must all be on the current Hardware Compatibility
List (HCL).
This requirement will be different for each of the operating system versions,
and can require verification and support by an OEM as with Windows 2000
Datacenter Server.
If your design requires functions not implemented as standard Microsoft

product, you must ensure that the vendor has had the hardware and software
tested and supports the Microsoft logo and HCL requirements.
When selecting hardware platforms it is recommended that your systems have
identical configurations for each functional use to avoid errors due to
incompatibilities between devices. When components are added to a hardware
platform, it is also recommended that the components be identical. For
example, if disk controllers are added to a hardware platform, use the same

model and revision of controller to ensure that the same driver can be used.
After completing this lesson, you will be able to:
!
Define a target availability level.
!
Identify highly available system and computer components.
!
Identify the appropriate use of teamed adapters for a highly available Web
infrastructure.
!
Select an appropriate hardware platform for the Web infrastructure.
!
Identify the appropriate redundant array of independent disks (RAID) levels
for a highly available Web infrastructure.
!
Identify highly available hardware components for Microsoft server cluster
and Network Load Balancing.

Introduction
Lesson objectives
Module 2: Designing Highly Available Web Infrastructure Servers 3


Required Level of Availability
!
Planned system downtime
!
Unplanned system downtime
!
Calculated values

!
Document required availability

*****************************
ILLEGAL FOR NON
-
TRAINER USE
******************************
When choosing a hardware platform, you must first determine the required
level of availability for each server in your Web infrastructure. Meeting
availability requirements must be part of both your design considerations and
your specifications for maintenance of the Web infrastructure.
You can maximize the availability of your Web solution by identifying and
minimizing single points of failure and by planning for system downtime.
System downtime falls into two major categories: planned and unplanned.
Planned downtime is anticipated and scheduled to have a minimum impact on
system availability. During planned downtime, system maintenance, re-
configuration, backup, and upgrade occurs.
Unplanned downtime is expected, but it is only acceptable if it is a low
percentage of the total resource downtime.
Unplanned downtime is the result of faults that occur in hardware or software
components. You can minimize unplanned downtime by using highly reliable
components or by designing solutions to include fault tolerance for both
software and hardware components.
Introduction
Planned system
downtime
Unplanned system
downtime
4 Module 2: Designing Highly Available Web Infrastructure Servers

When you have calculated values for the mean time between failures (MTBF)
and the mean time to repair (MTTR), you can predict availability for the system
by using the following formula:
System Availability% = (MTBF / (MTBF + MTTR)) * 100
By using the MTBF and MTTR value for a server that provides a particular
service, you can calculate availability. This calculated value is the amount of
time that it would take to repair the server and bring the server back online.
You can also determine availability by using service uptime, which is the
measured value for a server providing a particular service, and the service
downtime, which is the total time to repair and bring the service back to
operation. By using these values, you can use the following formula to calculate
the actual availability achieved.
Service Availability = Service uptime / (Service uptime + Service downtime)
You must document the required availability (number of nines) for each server
in your solution; this number may or may not be the same as the overall
availability required, depending on the design.
Your solution will most likely consist of many servers, each of which provides
different services; therefore, the availability requirement for each server and its
associated services may be different from the overall availability requirement
for your solution. For example, a Web-based e-commerce solution that allows
customers to view catalogs and place orders may require high availability for
the Web servers, because loss of service may render the site inaccessible.
Some servers in your solution may not be required to have extremely high
availability because the loss of the service provided would not prevent client
access to the solution. For example, the loss of a DHCP server, although critical
to the management network interfaces, may be tolerable for several hours if
address requests occur only when computers have their power cycled.
Calculated values

Document required
availability
Module 2: Designing Highly Available Web Infrastructure Servers 5


Highly Available System Components
!
PCI
!
I/O Cages
!
Host Bus Adapters
!
Network Cards
!
Power Supply
!
Fans
!
Motherboard
!
Drives
!
Memory

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

When designing a highly available Web infrastructure, choose components that
make a physical server highly available. The components that make up a
physical server are the same regardless of whether the physical server is built
for availability. However, a physical server built for high availability has
redundant and hot-pluggable components.
The following table lists system components and the considerations that you
must apply when making design decisions for your highly available Web
infrastructure.
System component Considerations

PCI A hot plug Peripheral Component Interconnect (PCI) bus that
allows you to add, remove, replace, and upgrade PCI controllers
without shutting down the system.
I/O cages A hot swap input/output (I/O) cage that allows you to add,
remove, replace, and upgrade PCI cards without shutting down
the system.
Host bus adapters Failover small computer system interface (SCSI) Host Bus
Adapters that merge matched SCSI controllers into controller
pairs that provide controller redundancy.
Failover Fiber Channel Host Bus Adapters that provide a
separate channel to the external storage array.
Network cards Redundant network adapter teaming.

Introduction
Hardware
considerations
6 Module 2: Designing Highly Available Web Infrastructure Servers


(continued)

System component Considerations

Power supply Redundant hot-plug power supplies that can be added or
replaced without shutting down the server.
N+1 balancing. N equals the number of power supplies that are
needed to provide enough power to the server efficiently. 1
equals the added power supply in case any of the n power
supplies fail.
Redundant processor power modules. Redundant processor
power modules prevent failure and monitor power currents to
detect and prevent potential problems proactively.
Fans Redundant fans that ensure proper airflow around temperature-
sensitive components if a fan fails.
Hot-plug fans that you can replace without shutting the system
down.
N+1 balancing. N equals the number of fans that are needed to
cool the server efficiently. 1 equals the added fan in case any of
the n fans fail.
Fans with speed sensors allow alarms to be raised if the fan
speed is out of tolerance.
Motherboard Offline processor recovery, which restarts the computer when a
processor fails. If a processor fails, the computer can detect it.
After the computer is restarted, it will not use that processor.
Advanced basic input/output system (BIOS) features can
support system alarms, such as fan speed and onboard voltage
errors.
Drives Hot-plug drives that permit you to plug and unplug SCSI drives
from the system while the system is in operation.
Memory Error-correction coding (ECC) memory, which enables
detection and correction of all single-bit memory errors.

Module 2: Designing Highly Available Web Infrastructure Servers 7


Highly Available Computer Components
!
Disk Controllers
!
Disk Drives
!
I/O Adapters
!
Network Adapters
!
Battery Backup Units

*****************************
ILLEGAL FOR NON
-
TRAINER USE
******************************
The most basic way to increase availability is to increase the reliability of your
hardware components; therefore, choose components that are either highly
available or support highly available configurations. For any components that
represent single points of failure in your system, choose hardware components
that have a high MTBF to provide the best possible reliability.
Reliability is the measure of how prone a hardware component is to failure. All
components and systems will eventually fail. A system’s failure rate is the
combined failure rates of its component parts. For this reason, you can increase
reliability and potentially achieve higher availability by using fewer, more

powerful components. By using this strategy, you can make many high-end
production systems more reliable (and therefore have a lower failure rate).
Introduction
Reliability of the
hardware components
8 Module 2: Designing Highly Available Web Infrastructure Servers


To ensure that you have highly available hardware components in your design,
consider the best practices listed in the following table.
Computer component Best practices

Disk controllers Choose controllers with ECC and battery backup cache
random access memory (RAM).
Choose controllers with parity or ECC checked I/O buses.
Choose fiber-based controllers whenever possible (fiber
channel).
Disk drives Use fiber channel whenever possible.
Choose drives that can be automatically assigned ID and
logical unit numbers to avoid manual configuration.
I/O adapters or other
peripherals
Ensure that adapters and peripherals are specified on the
appropriate HCL for the version of Windows 2000 that you
are using. For example, Windows 2000 Datacenter Server has
a separate HCL for components used in a validated server
running Windows 2000 Datacenter Server.
Ensure that I/O controllers are Plug and Play compliant to
ensure maximum flexibility in allocating I/O port and
interrupt resources.

Ensure I/O controllers have either signed built-in Microsoft-
supplied drivers or signed drivers to ensure that adequate
testing and verification has been performed for the version of
the operating system that you are using.
Network adapters Choose single port cards where I/O chassis supports hot swap
repair.
Choose multiport cards where hot swap is not possible.
Choose adapters that the driver configures (no separate
configuration utility is required).
Use the same model and version of network adapters
whenever possible and always when your system requires
teaming.
Battery backup units Choose controllers and computers that include battery backup
capabilities or provide external battery backup units to
minimize loss of data due to AC power loss.
For large datacenter environments, ensure that sufficient
instantaneous battery backup power is available to support a
transition to standby generators.

Best practices
Module 2: Designing Highly Available Web Infrastructure Servers 9


Network Configuration
!
Ensure that the server is capable of handling the
aggregate throughput of the cards to obtain the best
possible performance
!
Select single port adapters to allow maximum flexibility

in the configuration for high availability servers that
support many PCI slots with hot swapping capability
!
Ensure that the team members are on different PCI
bridges within the computer I/O structure, when
teaming network adapters
!
Choose network adapters for servers where the
number of PCI slots is limited and hot swap capabilities
are not supported

*****************************
ILLEGAL FOR NON
-
TRAINER USE
******************************
By selecting an appropriate network adapter, you can improve the availability
of data traveling on the local area network (LAN). Select a network adapter that
allows configurations that minimize the potential for attack and maximize
potential server performance.
Teaming Network adapters can be configured to provide load balancing and
fault tolerance. For example, Intel provides Pro/100S server adapters, which
support two to eight adapters that are configured to work together as an
Adaptive Load Balancing or Adapter Fault Tolerance team.
!
Adaptive Load Balancing. Allows balancing of the transmission data flow
among two to eight adapters and also includes the Adapter Fault Tolerance
option. Adaptive Load Balancing works with any gigabit switch. All
adapters must run at same speed.

Introduction
Teaming Network
adapters
10 Module 2: Designing Highly Available Web Infrastructure Servers


!
Adapter Fault Tolerance. Provides automatic redundancy for your adapter.
If the primary adapter fails, the secondary adapter takes over. Adapter Fault
Tolerance supports two to eight adapters per team with any combination of
speeds.



When your design requires the highest level of availability in the server,
consider mandating support for hot-swappable PCI cards, to that you can
change failed adapters without restarting the server.

If multiple network adapters are part of your hardware component design,
consider the following best practices:
!
Ensure that the server is capable of handling the aggregate throughput of the
cards to obtain the best possible performance. When configuring teams, it is
recommended that you balance the adapters evenly across all of the PCI
buses in the server.
!
Select single port adapters to allow maximum flexibility in the configuration
for high availability servers that support many PCI slots with hot swapping
capability. Where multiple adapters are used for a single logical connection,
teaming and failover configurations can be used to provide redundancy and

fault tolerance.
!
Ensure that the team members are on different PCI bridges in the computer
I/O structure when you team network adapters. If a member of the team
fails, it can potentially be hot swapped without disturbing other members of
the team.
!
Choose multiport network adapters for servers where the number of PCI
slots is limited and hot swap capabilities are not supported. Failures in
multiport adapters may result in total failure of the PCI card, affecting all
ports on the card or may result in downtime for the server when changing
the card.

Note
Best practices
Module 2: Designing Highly Available Web Infrastructure Servers 11


Highly Available Computer Hardware
!
Server architecture
!
Fans
!
PCI hot plug
!
Power supply needs
!
Random access memory

*****************************
ILLEGAL FOR NON
-
TRAINER USE
******************************
When designing your Web infrastructure, you must ensure that your computer
hardware platform is highly available. Consider that each computer in your
design may be required to meet different reliability needs.
You can use Network Load Balancing to make your computer hardware more
highly available by load balancing the incoming TCP/IP traffic across all of the
hosts in the cluster.
For computers that are already designed for high availability, you can design
your Web infrastructure to include server clusters. Server clusters preserve
client access to applications and resources during failures and planned outages,
therefore increasing availability levels.
If your solution requires stand-alone computer hardware, you must evaluate and
choose internal components based on the appropriate reliability levels for your
hardware platform design.
Your system design must address the need for greater bandwidth in the key
subsystems, including memory, I/O, and processors. You can use dual memory
controllers and dual-peer PCI buses that transfer data between each processor
and critical subsystems in parallel. ECC also protects the data being transmitted
between devices to guarantee the data integrity.
Many computer manufacturers separate their computers designed as servers
between two architectures:
!
Workgroup server architecture. Several workgroup servers in the x86
market support two processors to process instructions concurrently.
However, the overall system bandwidth of these servers is limited because
each processor must compete with the other for access to critical memory

and I/O subsystems.
!
Highly parallel system architecture. This architecture uses two PCI buses
and two memory controllers in parallel. This configuration significantly
increases the performance of multiprocessor systems.

Introduction
Server architecture