Tài liệu Module 3: Preparing for Cluster Service Installation docx

Contents Overview 1 Pre-Installation Requirements 2 Identifying Hardware Considerations 8 Assigning IP Addresses Within a Cluster 16 Assigning Names Within a Cluster 18 Determining

Contents

Overview 1

Pre-Installation Requirements 2

Identifying Hardware Considerations 8

Assigning IP Addresses Within a Cluster 16

Assigning Names Within a Cluster 18

Determining Domain Considerations 20

Existing Services and Applications 23

Lab A: Configuring Advanced Server for

to represent any real individual, company, product, or event, unless otherwise noted Complying with all applicable copyright laws is the responsibility of the user No part of this document may

be reproduced or transmitted in any form or by any means, electronic or mechanical, for any purpose, without the express written permission of Microsoft Corporation If, however, your only means of access is electronic, permission to print one copy is hereby granted

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

 2000 Microsoft Corporation All rights reserved

Microsoft, Active Directory, BackOffice, Jscript, PowerPoint, Visual Basic, Visual Studio, Win32, Windows, Windows NT are either registered trademarks or trademarks of Microsoft Corporation

in the U.S.A and/or other countries

Other product and company names mentioned herein may be the trademarks of their respective owners

Program Manager: Don Thompson

Product Manager: Greg Bulette

Instructional Designers: April Andrien, Priscilla Johnston, Diana Jahrling

Subject Matter Experts: Jack Creasey, Jeff Johnson

Technical Contributor: James Cochran

Classroom Automation: Lorrin Smith-Bates

Graphic Designer: Andrea Heuston (Artitudes Layout & Design)

Editing Manager: Lynette Skinner

Editor: Elizabeth Reese

Copy Editor: Bill Jones (S&T Consulting)

Production Manager: Miracle Davis

Build Manager: Julie Challenger

Print Production: Irene Barnett (S&T Consulting)

CD Production: Eric Wagoner

Test Manager: Eric R Myers

Test Lead: Robertson Lee (Volt Technical)

Creative Director: David Mahlmann

Media Consultation: Scott Serna

Illustration: Andrea Heuston (Artitudes Layout & Design)

Localization Manager: Rick Terek

Operations Coordinator: John Williams

Manufacturing Support: Laura King; Kathy Hershey

Lead Product Manager, Release Management: Bo Galford

Lead Technology Manager: Sid Benavente

Lead Product Manager, Content Development: Ken Rosen

Group Manager, Courseware Infrastructure: David Bramble

Group Product Manager, Content Development: Julie Truax

Director, Training & Certification Courseware Development: Dean Murray

General Manager: Robert Stewart

Instructor Notes

This module covers the information that is required to plan for the installation

of Microsoft® Cluster service Requirements specific to server clusters include communication networks, shared disks, and data storage, in addition to hardware considerations for the operating system This module describes the naming and addressing conventions that the cluster components require

Domain considerations are covered, in addition to issues relating to services and

applications that were installed and running prior to installing Cluster service

After completing this module, students will be able to:

Determine the pre-installation considerations for a server cluster

Identify hardware considerations

Assign Internet Protocol (IP) addresses within a cluster

Assign names within a cluster

Determine domain considerations

Determine pre-installation considerations for existing services and applications

Materials and Preparation

This section provides the materials and preparation tasks that you need to teach this module

Required Materials

To teach this module, you need the following materials:

Microsoft PowerPoint® file 2087A_02.ppt

Access to the Hardware Compatibility List at www.microsoft.com/hcl

Preparation Tasks

To prepare for this module, you should:

Read the materials for this module and anticipate questions students may ask

Be familiar with the HCL

Be familiar with the Fibre Channel technologies from the leading hardware vendors

Practice the lab

Be familiar with the gratuitous Address Resolution Protocol (ARP) procedure that Cluster service generates after a failover, and how that relates

to routers and switches

Study the review questions and prepare alternative answers for discussion

Presentation:

60 Minutes

Lab:

15 Minutes

Module Strategy

Use the following strategy to present this module:

Students need to be aware of pre-installation requirements and considerations that will enable them to successfully install and deploy Cluster service

Pre-Installation Requirements There is information regarding pre-installation tasks on the opening page of this section Be sure students are aware of the need to back up data and determine potential points of failure Additionally, students need to know about the special requirements of Cluster service for network connections, cluster disks (shared disks), and data storage

• Cluster Network Requirements: Students need to be able to identify the public and private network methodology and what travels across those different networks and how they relate to a server cluster

• Cluster Disk Requirements: Prior to the installation of Cluster service, the student must be able to verify that the server that the cluster will use meets the disk requirements Students need to configure the disks prior

to installation of Cluster service Students can add disks to the cluster after they install the cluster, and the disks must meet the same requirements

• Data Storage Requirements: Stress to the students that the operating system is on the local drives and that all of the partitions of the shared disks must have identical drive letters on each node of the cluster The local drives and partitions of the shared disk must be set up before installation

Identifying Hardware Considerations Cluster service is more dependent on hardware than other software produced by Microsoft Because this dependency is such a critical issue, the hardware used should be on the HCL Students also need to consider hardware outside of the server cluster and how it may affect the performance of the cluster

• Cluster Service Compatible Systems: Each node of a cluster requires a minimum set of hardware for Cluster service to install properly Stress to the students that this is an Enterprise solution and that the hardware should not be substandard

• Routers, Switches, and Hubs: When a resource fails over, the node controlling the failed over group must send a gratuitous ARP update to let clients, switches, and routers know about the ARP update Some legacy switches and routers might not be able to accept this ARP update

or forward it to clients Students need to be aware that there are more things controlling the outcome of a successful deployment besides the use of two servers and the software they are running

• Network Cards: Cluster service does not require separate network cards for public and private communication, but you need to stress to students that it is highly recommended Using a single adapter for public and private communication can result in failovers of resources that are caused by traffic congestion on this single network Remind the students that there are many types of supported network cards, such as Ethernet, Fibre, and other specialized interconnects

• Cluster Disks: Disks that are on the same bus between two servers are shared disks After Cluster service is installed, these disks will be referred to as cluster disks Considerations for these disks focus on throughput For instance, if disk access is a concern, you can use a faster disk or implement hardware Redundant Array of Independent Disks (RAID), or move the disk to another shared bus between the two nodes

of a cluster if there is contention for the input/output (I/O) of the shared bus

• Cluster Data Access: A server cluster can use a disk that is accessed through a small computer system interface (SCSI) or Fibre bus Most server cluster implementations use Fibre

Assigning IP Addresses Within a Cluster: Cluster service needs a minimum

of five static IP addresses Remind the students that although the private network can use automatic private addressing, which is a feature of Microsoft Windows® 2000, the servers will have faster startup times with a static address for the private network

Assigning Names Within a Cluster: Cluster service uses network basic input/output system (NetBIOS) names for the administration of the cluster, virtual servers, and for nodes of the cluster Inform the student that as a best practice, they should use the cluster name for administration only, and they should treat virtual server names as a separate group of resources

Determining Domain Considerations

• In this section you will talk about how server clusters interact with a Windows 2000 domain, and what accounts students will need to create

or manage prior to installing Cluster service

• User Accounts: The two user accounts that are of concern prior to installation of server cluster are the account used to install the service, which needs to have administrative rights on each node of the cluster, and the service account, which also has to have administrative rights and the right to log on as a service

• Computer Accounts: The nodes have computer accounts in the domain

to which they belong, whether they are domain controllers or member servers It is a best practice to keep the member server computer accounts in their own organizational unit so that group policies do not affect them

Existing Services and Applications

• If you are upgrading a server to become a node in a cluster, you need to consider whether the services and applications that are running on the server will work after the installation of Cluster service For example, Windows Internet Name Service (WINS) can continue to run outside the cluster on a server that is installed as a node Microsoft Exchange, however, will not run on a server that is upgraded to a server cluster environment and will fail Tell the students that extensive testing of applications and services needs to be done if the applications and services are running on a server that will be installed with Cluster service

Instructor Setup for a Lab

Lab Strategy

This lab is designed to prepare the servers for the installation of Cluster service

In this lab the students will set up identical drive letters on each server in the cluster Then they will configure the private and public networks for cluster communications At the end of the second exercise, there is a closing exercise using the Pre-Installation Checklist This checklist will also serve as a job aid for students to use later on

Lab A: Configuring Advanced Server for Cluster Installation

To conduct this lab:

Read though the lab carefully, paying close attention to the instructions and details

Students work in teams of two, grouped together by their shared bus

Help the students determine whether they are Node A or Node B In these exercises, students will perform all of the steps on both servers

Familiarize the students with the Reference Table and how to find their computers in the table

Allow some time to discuss the Pre-Installation Checklist

Overview

Pre-Installation Considerations

Identifying Hardware Considerations

Assigning IP Addresses Within a Cluster

Assigning Names Within a Cluster

Determining Domain Considerations

Existing Services and Applications

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

This module covers the information that is required to plan for the installation

of Microsoft® Cluster service Requirements specific to server clusters include communication networks, shared disks, and data storage, in addition to hardware considerations for the operating system This module describes the naming and addressing conventions that the cluster components require

Domain considerations are covered, in addition to issues relating to services and applications that were installed and running prior to installing Cluster service After completing this module, you will be able to:

Determine the pre-installation requirements for a server cluster

Identify hardware considerations

Assign Internet Protocol (IP) addresses within a cluster

Assign names within a cluster

Determine domain considerations

Determine pre-installation considerations for existing services and applications

This module outlines what

you need to do before

installing Cluster service

Power

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

Before installing Cluster service you will need to plan for failures that could occur in the environment You also need to consider how the nodes in the cluster will communicate with each other, and how clients will access the server cluster Additionally, you must ensure that your system will have the shared disks and data storage that are required for a successful Cluster service installation

Backup and Restore

Clustering technology provides increased reliability; however, it is not designed

to protect all of the components of your workflow in all circumstances For example, Cluster service is not an alternative to backing up data, because it protects only the availability of data, not the data itself Therefore, you need to plan for backup and restoration of data

Risk Assessment

It is recommended that prior to creating a cluster, you complete a risk assessment of your network to identify possible single point of failures that can

interrupt access to network resources A single point of failure is any

component in the environment that would block client access to data or applications if it failed Single points of failure can be hardware, software, or external dependencies Examples of single points of failure include dedicated wide area network (WAN) lines or the power supply from a utility company

Uninterruptible Power Supply

Another recommendation is to consider providing an uninterruptible power supply (UPS) protection for individual computers and the network, including hubs, bridges, and routers A UPS is just one more factor in configuring a total fault tolerant solution Many UPS solutions provide power for 5 to 20 minutes, which is long enough for the Microsoft Windows® 2000 operating system to properly shut down when power fails

Topic Objective

To determine the

pre-installation considerations

for a server cluster that will

protect the components and

data in case of a failure

Lead-in

A server cluster does not

address single points of

failure outside of the cluster

You need to assess the

current environment before

you install a server cluster

Cluster Network Requirements

Public Network

Private Network

Mixed Network

Private Network

Public or Mixed Network

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

A cluster network has two primary types of communication The private communication provides online status and other cluster information to the nodes The public communication provides information between the client and virtual servers It is recommended that private network communications be physically separated from public network communications, but have the ability

to use the mixed network to eliminate a single point of failure

An alternative network type a mixed network, uses the public network for both

private and public communications The advantage of a mixed network is that private network communications can be failed over to the public network If you dedicate a network to client-to-node communications, node-to-node communications cannot fail over to that network The mixed network configuration is the preferred configuration for a public network

Network adapters, known to the cluster as network interfaces, attach nodes to networks You configure what types of communication will travel across the networks by using the cluster administration tools

Topic Objective

To describe how Cluster

service uses public, private,

and mixed networks for

cluster and client

communications

Lead-in

Server clusters use a

network for

communications You have

three configuration options

for server cluster

communications

Delivery Tip

Stress to the students that a

dedicated public network will

never transmit private

communications – have at

least one mixed network

Private Network

The private network, also known as the interconnect in a server cluster, will potentially carry the following five types of information:

Server heartbeats Each node in a cluster exchanges IP packets

(approximately every 1.2 seconds) with the other node in the cluster determining if both nodes are running correctly The first node of the cluster

to come online is the node that is responsible for sending out the heartbeats The second node begins to send heartbeats to inform the first node that the second node has come online If a node does not respond to the heartbeat, the working node identifies the unresponsive node as having failed If a node fails to detect six successive heartbeats from another node, the other node assumes that the unresponsive node is offline and tries to take ownership of the resources that the nonresponding node owns Note that failure to detect a heartbeat message can be caused by node failure, network interface failure, or network failure

Replicated state information Every node in the cluster uses replicated state

information to communicate which cluster groups and resources are running

on all of the other nodes

Cluster commands Cluster commands manage and change the cluster

configuration An example of a cluster command would be any node communications regarding adding or removing a resource or failing over a group

node-to-
Application commands Cluster-aware applications send application

commands through the interconnect to communicate with copies of an application running on multiple servers

Application data Application data is when cluster-aware applications share

data between nodes

If the private network fails over to the public network, the Cluster service employs packet signing for node-to-node communications to provide additional security over the network

Public Network

The public network connection extends beyond the cluster nodes and is used for client-to-cluster communication The public network cannot function as a backup connection for node-to-node communication should the private network fail The network interface cards (NICs) for the public network must be on the same subnet

Note

Mixed Network

The typical server cluster will have one NIC on each server that is designated for internal communications (cluster only), and one or more other NICs designated for all communications (the mixed network serving both cluster and client) In such a case, the cluster-only NIC is the primary interconnect, and the mixed network NIC(s) serves as a backup if the primary ever fails

Cluster service can only do this for the primary cluster interconnect That is, it provides the ability to use an alternate network for the cluster interconnect if the primary network fails This eliminates an interconnect NIC from being a single point of failure There are vendors who offer fault tolerant NICs for

Windows 2000 Advanced Server, and you can use these for the NICs that connect the servers to the client network

Cluster Disk Requirements

All Disks on Shared Bus

Disks can be seen from all nodes

Basic Disks, not Dynamic

All Disk Partitions must be NTFS

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

A shared disk is a resource that all of the nodes in a cluster can access This shared disk will be known as the cluster disk after installation of Cluster service The cluster disk allows clients to access data even if a node goes offline To install Cluster service on Windows 2000 with a properly functioning cluster disk, you will need to make certain that:

All of the shared disks, including the quorum disk, are physically attached

to a shared bus or buses

Disks attached to the shared bus can be seen from all of the nodes You can verify that you can see all of the nodes at the host adapter setup level

Shared disks are configured as basic, not dynamic

All disks must be formatted NTFS

Cluster service supports hardware RAID not software RAID

While not required, the use of fault tolerant hardware Redundant Array

of Independent Disks (RAID) is strongly recommended for all cluster disks

Topic Objective

To describe the function and

requirements of the shared

disk which both nodes of a

server cluster access

Lead-in

For a cluster to use a

shared disk, the disk must

meet the following criteria

Note

Data Storage Requirements

C: = Local Disk 0 W: = Cluster Disk 1 X: = Cluster Disk 1 Y: = Cluster Disk 2

Node B Disk Configuration:

C: = Local Disk 0 D: = Local Disk 1 W: = Cluster Disk 1 X: = Cluster Disk 1 Y: = Cluster Disk 2

W:

Node A

Local Disk 0

Local Disk 1

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

Cluster service is based on the shared nothing model of clustering In a shared

nothing model, only one node in the cluster can logically control a disk resource

at a time Having only one node controlling a disk resource at any given time allows the Windows 2000 cluster file system to support the native NTFS file system rather than requiring a dedicated cluster-aware file system

Identical Drive Letters

After you have configured the bus, disks, and partitions, you must assign drive letters to each partition on each clustered disk Drive letters for cluster disks and partitions must be identical on both nodes For example, Microsoft

SQL Server™ is running on a virtual server that Node A controls with the SQL database on drive W If the virtual server fails over to Node B, SQL server can access data on Node B’s drive W

It is usually best to assign drive letters from the end of the alphabet to avoid duplicate drive letters

Every partition’s drive letter

on the cluster disks must be

the same for both nodes in

the cluster

Note

Identifying Hardware Considerations

Selecting a Cluster Service Compatible System

Routers, Switches, and Hubs

Network Cards

Cluster Disk

Cluster Data Access

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

You can choose from a wide range of cluster compatible systems that meet the minimum Cluster service requirements It is recommended that all of the hardware be identical for both nodes Using identical hardware makes configuration easier and eliminates potential compatibility problems Microsoft supports only complete cluster configurations that have passed cluster

validation testing Select the Cluster category in the Hardware Compatibility

List (HCL) at http://www.microsoft.com/hcl/default.asp to view the list of tested and validated configurations

The HCL also lists various system components, such as small computer system interface (SCSI) adapters, Fibre Channel adapters, and RAID devices, that have passed cluster component candidate testing

Using hardware that is incompatible with Cluster service can result in

a variety of problems, including failure of the nodes to start or to restart It is strongly recommended that you use hardware that is on the Windows 2000 Hardware Compatibility List for Cluster service

A server cluster is more

reliant on hardware than

other software applications

You need to identify the

appropriate hardware to use

for a successful cluster

implementation

Delivery Tip

Emphasize the message in

the Warning note

If you have Internet access,

open the HCL Web site

Warning

Cluster Service Compatible Systems

Dual PCI Bus

PCI Disk Controller

Separate Disk Controller for the shared bus

PCI Network Adapter

External Hard Disk

Cables and Terminators

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

Each node in a cluster requires all of the hardware requirements for Windows 2000 Advanced Server, plus the following requirements that are specific to Cluster service:

A Dual Peripheral Component Interconnect (PCI) bus

A PCI disk controller

A separate disk controller from the operating system that must be used for the shared data bus

A minimum of one PCI network adapter ISA network cards are not suitable for use in a cluster because of slow throughput Saturation on an ISA card can lead to delays in packet transmission for up to ten seconds Cluster service will wait for five seconds before

determining failover of the other node

An external hard disk that is connected to both computers that the cluster will use as the shared disk

Cables and terminators to attach the disk to both computers and properly terminate the bus

The Cluster Verification utility from www.microsoft.com will verify proper installation of a two-node cluster However, the tests that are performed

on the cluster will destroy all data on the clustered disks Do not use this utility with an established cluster

Topic Objective

To list the hardware

requirements for Cluster

service

Lead-in

A server cluster installation

will consist of the following

hardware

Note

Note

Routers, Switches, and Hubs

\\Accounting

IP 10.0.0.5

Router

IP 192.168.5.1 MAC 00-D8-60-33-FA

NodeA

IP 10.0.0.1 MAC 00-40-96-32-37-BA

NodeB

IP 10.0.0.2 MAC 00-D0-59-12-0F-00

\\Accounting=

IP 10.0.0.5 MAC 00-40-96-32-37-BA

\\Accounting=

IP 10.0.0.5 MAC 00-D8-60-33-FA

\\Accounting=

IP 10.0.0.5 MAC 00-40-96-32-37-BA

\\Accounting

IP 10.0.0.5

ARP Update

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

A server cluster installation requires hardware compatible routers, switches, and hubs For the client to respond to a failed virtual server, Microsoft Cluster service sends a gratuitous Address Resolution Protocol (ARP) update to the network devices The ARP update notifies the client of the new media access control (MAC) to IP address association The clients can then redirect client to virtual server communication to the new controlling node

You need to verify that hubs and switches can forward the ARP update to clients and routers However, some devices, such as switches, may not forward the gratuitous ARP request to other devices Hubs will always forward ARP updates A switch may not forward the update, but you can configure it to forward the updates A router never forwards the update,\; however it needs to

be able to accept the update and change its ARP table It is important to choose and test routers and switches for compatibility prior to implementation of a server cluster

In this example, Node A is controlling \\Accounting with an IP address of 10.0.0.5 The clients and routers contain an address table of the IP to MAC relationships When the virtual server changes ownership, Node B sends out a gratuitous ARP update to the local subnet Hubs and switches forward the information to clients and routers on the local subnet Routers contain an IP to MAC address table on behalf of clients on remote subnets, therefore routers do not forward the information but must be able to accept the ARP update

Topic Objective

To identify considerations

for ensuring that routers,

switches, and hubs in a

server cluster are

compatible with Cluster

service

Lead-in

Networking devices must be

able to accept an ARP

update after a group

failover

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

The typical server cluster has one NIC on each server which is designated for internal, cluster-only communications One or more other NICs are designated for all public communications, including a mixed network that can serve both cluster and client

Network Descriptions

If the cluster node uses multiple, identical PCI network cards, it may be difficult to identify them when you run the Cluster service setup Microsoft Windows 2000 setup assigns network descriptions to each card You will need

to identify which network cards that the cluster will use for private, public, or mixed cluster networks, and enter descriptions for each

For example, you would use the Transmission Control Protocol/Internet Protocol (TCP/IP) Ipconfig utility to display the network driver name with an index (such as E190x1 and E190x2) and the network card’s IP address and subnet mask Using this information, you can then assign appropriate names to the networks when you run the Cluster service setup For example, if El90x1 uses a private IP address 10.0.0.1, this address is for node-to-node

communication If El90x2 uses an IP address on the public network, this address is for client-to-node communication

Two network adapters are recommended so that the nodes of the cluster can have a private network for node-to-node communications

Topic Objective

To explain the use of

network cards in a server

Supported Network Types

A supported Cluster service configuration can use as its interconnect virtually any network technology that Windows 2000 Advanced Server supports This includes the following:

10BaseT Ethernet

100BaseT Ethernet

Fiber Distributed Data Interface (FDDI) network card

Specialized interconnect technologies such as Tandem ServerNet and GigaNet Cluster LAN (cLAN)

Cluster Disks

Hardware RAID

Disk Access Speed

Multiple Shared Bus

Hardware RAID Single Disk

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

Applications and services access data that is stored on the cluster disk In a multiple server environment, throughput may become a concern on the shared bus You need to consider the demand for data from the applications and services on the cluster to decide whether to implement the following cluster disk solutions

Hardware RAID

You could configure hardware RAID on the external disk device Hardware RAID will increase data access on the cluster disk for greater performance A

stripe set, a feature of RAID, will also provide faster read/write functionality

for data access

Disk Access

Another alternative is to select a faster disk, realizing that you will then need a wider bus A faster disk will provide increased read/write functions for data access For example, you may select a wide SCSI with a maximum transfer rate

of 20 megabytes (MB) per second, or you may select an ultra-wide SCSI with a maximum transfer rate of 40 MB per second

Multiple Shared Bus

If you have a number of disks on the same, shared bus, you may find a decrease

in performance of read/write data access In an active/active cluster environment, both nodes access disks on the same shared bus To increase read/write performance, consider moving some disks to a separate shared bus

Cluster service does not support the use of software RAID for the cluster disks

Topic Objective

To identify possible

problems and solutions with

throughput because server

cluster applications and

services are on the cluster

disk

Lead-in

Cluster disks could have

contention of the shared bus

from different nodes on the

cluster You can improve

performance by

implementing RAID, a faster

disk, or another bus

Note