Xem mẫu

  1. Module 3: Preparing for Cluster Service Installation 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 Cluster Installation 24 Review 31
  2. Information in this document is subject to change without notice. The names of companies, products, people, characters, and/or data mentioned herein are fictitious and are in no way intended 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
  3. Module 3: Preparing for Cluster Service Installation iii Instructor Notes Presentation: This module covers the information that is required to plan for the installation 60 Minutes of Microsoft® Cluster service. Requirements specific to server clusters include communication networks, shared disks, and data storage, in addition to Lab: hardware considerations for the operating system. This module describes the 15 Minutes 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.
  4. iv Module 3: Preparing for Cluster Service Installation 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.
  5. Module 3: Preparing for Cluster Service Installation v • 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.
  6. vi Module 3: Preparing for Cluster Service Installation 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.
  7. Module 3: Preparing for Cluster Service Installation 1 Overview Topic Objective To provide an overview of the module topics and Pre-Installation Considerations objectives. Lead-in Identifying Hardware Considerations This module outlines what you need to do before Assigning IP Addresses Within a Cluster installing Cluster service. 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.
  8. 2 Module 3: Preparing for Cluster Service Installation Pre-Installation Requirements Topic Objective To determine the pre- installation considerations for a server cluster that will Client Client protect the components and Client data in case of a failure. Client Client Client Lead-in A server cluster does not Client address single points of Client failure outside of the cluster. Router You need to assess the Router current environment before you install a server cluster. Server Server Power 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.
  9. Module 3: Preparing for Cluster Service Installation 3 Cluster Network Requirements Topic Objective To describe how Cluster service uses public, private, Private Public Network and mixed networks for Network cluster and client Private Network communications. Lead-in Mixed Network Server clusters use a network for communications. You have three configuration options for server cluster communications. Public or Mixed Network *****************************ILLEGAL FOR NON-TRAINER USE****************************** A cluster network has two primary types of communication. The private Delivery Tip communication provides online status and other cluster information to the Stress to the students that a nodes. The public communication provides information between the client and dedicated public network will virtual servers. It is recommended that private network communications be never transmit private physically separated from public network communications, but have the ability communications – have at to use the mixed network to eliminate a single point of failure. least one mixed network. 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.
  10. 4 Module 3: Preparing for Cluster Service Installation 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-to- node communications regarding adding or removing a resource or failing over a group. 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. Note 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.
  11. Module 3: Preparing for Cluster Service Installation 5 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.
  12. 6 Module 3: Preparing for Cluster Service Installation Cluster Disk Requirements Topic Objective To describe the function and requirements of the shared All Disks on Shared Bus disk which both nodes of a server cluster access. Disks can be seen from all nodes Lead-in For a cluster to use a Basic Disks, not Dynamic shared disk, the disk must All Disk Partitions must be NTFS meet the following criteria. *****************************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. Note While not required, the use of fault tolerant hardware Redundant Array of Independent Disks (RAID) is strongly recommended for all cluster disks.
  13. Module 3: Preparing for Cluster Service Installation 7 Data Storage Requirements Topic Objective Node A Disk To explain the data storage Configuration: requirements of a server Local Disk 0 Local Disk 0 C: = Local Disk 0 Node A cluster. Node A W: = Cluster Disk 1 C: X: = Cluster Disk 1 Lead-in Cluster Disk 1 Cluster Disk 1 W: Y: = Cluster Disk 2 Every partition’s drive letter 2 Partitions 2 Partitions X: on the cluster disks must be the same for both nodes in the cluster. Cluster Disk 2 Cluster Disk 2 Y: Node B Disk 1 Partition 1 Partition Local Disk 0 Configuration: Local Disk 0 C: = Local Disk 0 C: D: = Local Disk 1 Node B Node B W: = Cluster Disk 1 D: X: = Cluster Disk 1 Local Disk 1 Local Disk 1 Y: = Cluster Disk 2 *****************************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. Note It is usually best to assign drive letters from the end of the alphabet to avoid duplicate drive letters.
  14. 8 Module 3: Preparing for Cluster Service Installation Identifying Hardware Considerations Topic Objective To identify hardware and configuration considerations Selecting a Cluster Service Compatible System for a cluster. Lead-in Routers, Switches, and Hubs A server cluster is more reliant on hardware than Network Cards other software applications. Cluster Disk You need to identify the appropriate hardware to use Cluster Data Access for a successful cluster implementation. *****************************ILLEGAL FOR NON-TRAINER USE****************************** You can choose from a wide range of cluster compatible systems that meet the Delivery Tip minimum Cluster service requirements. It is recommended that all of the Emphasize the message in hardware be identical for both nodes. Using identical hardware makes the Warning note. configuration easier and eliminates potential compatibility problems. Microsoft supports only complete cluster configurations that have passed cluster If you have Internet access, validation testing. Select the Cluster category in the Hardware Compatibility open the HCL Web site. 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. Warning 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.
  15. Module 3: Preparing for Cluster Service Installation 9 Cluster Service Compatible Systems Topic Objective To list the hardware requirements for Cluster Dual PCI Bus service. Lead-in PCI Disk Controller A server cluster installation will consist of the following Separate Disk Controller for the shared bus hardware. 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 Note 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 Note 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.
  16. 10 Module 3: Preparing for Cluster Service Installation Routers, Switches, and Hubs Topic Objective \\Accounting IP NodeA To identify considerations IP \\Accounting= for ensuring that routers, MAC 00-40-96-32-37-BA IP switches, and hubs in a MAC 00-D8-60-33-FA server cluster are \\Accounting= IP compatible with Cluster MAC 00-40-96-32-37-BA service. Lead-in Networking devices must be able to accept an ARP update after a group failover. Router ARP IP Update MAC 00-D8-60-33-FA NodeB \\Accounting= IP IP \\Accounting MAC 00-D0-59-12-0F-00 MAC 00-40-96-32-37-BA IP *****************************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 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.
  17. Module 3: Preparing for Cluster Service Installation 11 Network Cards Topic Objective To explain the use of network cards in a server Separate Network Cards cluster. Lead-in Network Descriptions Server clusters use a variety of networks to communicate Supported Network Types with other nodes and with 10 BaseT clients. 100 BaseT FDDI Network Card Specialized Interconnects *****************************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, 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. Note Two network adapters are recommended so that the nodes of the cluster can have a private network for node-to-node communications.
  18. 12 Module 3: Preparing for Cluster Service Installation 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)
  19. Module 3: Preparing for Cluster Service Installation 13 Cluster Disks Topic Objective To identify possible problems and solutions with Hardware RAID throughput because server cluster applications and Disk Access Speed services are on the cluster disk. Multiple Shared Bus Lead-in Single Disk Hardware RAID 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. *****************************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. Note Cluster service does not support the use of software RAID for the cluster disks.
  20. 14 Module 3: Preparing for Cluster Service Installation Cluster Data Access Topic Objective To compare SCSI and Fibre benefits and requirements SCSI Requirements for Cluster Service for cluster data access. Lead-in Fibre Requirements for Cluster Service Server clusters can use SCSI or Fibre for the shared bus. Fibre is preferred because of greater I/O and configuration. *****************************ILLEGAL FOR NON-TRAINER USE****************************** Cluster service accesses data that is shared by either a SCSI bus or a Fibre Channel bus. Based on the amount of throughput that is needed for your cluster, you will need to decide which will meet your requirements. SCSI Fibre Channel Cost Lower cost Increased cost Configuration Difficult Easy Storage Area Network Not supported Supported (SAN) -capable Hardware Requirements No specialized equipment Specialized equipment Transfer Rate Performance 160 MB per second 266 MB per second Optical Cables N/A 10 km (maximum) Copper Cables 25 meters 100 meter
nguon tai.lieu . vn