12. Microsoft SQL Server and Microsoft Cluster Services
Type of trouble
MSCS Overview
Example of a cluster system
SQL Server Cluster Settings
Beyond MSCS
Summary of this chapter
In recent years, the Microsoft SQL Server system has been transferred from the desktop system and the working group to the enterprise backgroundhold system. In order to cope with business operations, these systems become more important, in other words, they must be more stable, easier to manage, and better fault tolerance. In order to meet these needs, Microsoft has bets considerable amounts of time and energy to reduce the errors of the software and improve the system's support. Microsoft also improves management tools, distal management capabilities, and develops new technologies like Microsoft Cluster Services (MSCs). Cluster refers to a computer that can be replenished in each other in a fault occurs. In this chapter, you will learn how MSCs work, how to set, and how to plan to recover old views from disasters. Also les on MSCs to make your system to make your system; you must cooperate with cautious planning to make your system have the ability to reply from errors.
Description
MSCs are included in Microsoft Windows 2000 Advanced Server, Microsoft Windows 2000 DataCenter Server, and Microsoft Windows NT 4.0 Enterprise Edition.
The type of fault is the main task of a library manager to maintain the database, and let the database can operate normally during the specified time period, which usually also puts it in the service scope consent. The service scope consent must also specify the best execution time, execution rate, and error during execution of the system must be provided, and will be restored during the time. Using MSCs can increase the longest execution time and reduce the time required to restore. Although the servo hard body, Windows 2000, Windows NT, and SQL Server generally say quite stable and reliable, some components sometimes have faults. In fact, a complex computer system may happen the following different types of faults: Although the disk technology has improved, the hard drive is still a mechanical device and will also be worn. Hard disc is one of the most often troubleshooting components. Hardware component faulty hardware components may be in charge of wear or destruction of components, causing such problems, which is inevitable in the system. Even computer devices that are best processes may be faulty. Soft components failure Some software will be discovered in a particular case. Before some specific conditions found the problem, the system may have already implemented a few months or have been in several years. In addition, the new application is also possible to modify key libraries or files in a stable environment, thus causing troubleshooting issues. External fault systems may also occur because some external factors have an error, and the power supply is suddenly interrupted. Do you have a way to live in such an error, you have to see if you have a Uninterruptible Power Supply, UPS or other remaining power.
The human error cluster is usually unable to protect the system that is faulty because of the human error. These people include accidentally deleting a data sheet or deleting a Windows NT archive system split area. Errors are inevitable, how to make the best preparation for these errors, is the focus of our discussion in this chapter. MSCS Overview MSCs is the built-in service of Microsoft Windows 2000 Advanced Server, Microsoft Windows 2000 Datacenter Server, and Microsoft Windows NT 4.0 Enterprise Edition. MSCS is used to construct a server cluster; as described above, the server cluster is a set of independent servo into a group and works together like a single system. The purpose of the cluster is to continue to access applications and other resources in order to occur during error events or expected possible interruptions. If a server is unable to use a server, the resource and application will move to other nodes in the cluster. When we talk about the cluster system, most of the high availability (high availability) is used instead of fault (FAULT TOLERANT). Traditionally, the fault tolerance is compared to a particular system, which provides high-order backup and recovery. Such systems typically use extremely professional software, allowing the system to instantly recover from any single hard body or soft error. The fault-tolerant system can be said to be quite expensive than the system that cannot be fault tolerant. Relative, a cluster system that provides high availability, is not as amazing like a fault-tolerant system. The cluster system is generally constructed by the criteria of the standard server hardware and the work system. When the availability requirements of the system are increasing, it can also be easily joined in the cluster to join the system as a node. Although a cluster system cannot guarantee that the continuity of continuity can be performed, the clustering system has indeed enhanced considerable availability for most of the application of most key task properties. MSCs have the following advantages: High availability system resources (such as disk or IP orders) automatically transfer from faulty servers to servers that provide services. This process is called falring. When an application in the context is faulty, MSCS automatically activates the application in another servo of another service, or dispersing the work of the faulty server to other nodes. The process of error-transfer is very fast, and the user aware of the service suspension of the server will be extremely short. Fault-tolerant reply When the faultful server is repaired and returned to the line, MSCS automatically re-balances the workload in the cluster. This is called false reply. The manageable cluster system administrator software allows you to manage the entire cluster as a single system. You can use the drag cluster object to move your application to different servers in the cluster in the cluster system administrator. You can also use the same way to move information. This "drag type" operation can be used to manually balance the load of the server or remove the load on a server to allow the server to operate. Cluster system administrators can also monitor the status of the cluster, each node, and all resources from any location on the Internet. Figure 12-1 shows the window of the cluster system administrator. Deliciousness When the needs of the system and load increase, you can reset the MSCS to support these needs. If the overall load exceeds the capacity of the cluster, you can also add new nodes in the cluster. Figure 12-1 Windows 2000 Cluster System Administrator
The basic concept of MSCS can reduce system shutdown time because MSCS can provide error-transfer functions between multiple systems. To reach the wrong transfer function, you must make the server interconnect, and use a shared disk SHARED Disk System, as shown in Figure 12-2. You can use any high-speed wiring system, such as an E. Bar, or other network hardware. The server interconnect system plays a communication channel between the server. This communication channel is to transfer the current status information of the cluster of the cluster, and configure information. The shared disk system can allow the database and data files of all servers in the bushes to be accessed in equal places. The shared disk system can be SCSI, Fiber Channel SCSI, or other exclusive hardware. As for a shared disk, it can be a separate disk or a RAID system (the RAID system has been discussed in Chapter 5). Note If the shared disk system does not have fault, and your disk subsystem fails, MSCs will be erroneous to move to other servers, but the new server will still use the wrong disk subsystem, so your system will still In a fault state. Since such a mechanical device is prone to failure, you must use RAID to protect your disc.
Once your system is set to a member of the cluster server, it will transform from the traditional server into a virtual server. Virtual Server looks like a normal server, but the system's entity identification has abstained. Because the computer hardware that makes up the virtual server may constantly change, the user can't recognize that the current application is actually executed in any time. Therefore, the virtual server provides services for the user's application, but it is not a specific hard body.
Figure 12-2 Windows 2000 Cluster
The virtual server exists on the Internet and also assigned a TCP / IP IP address. This IP address is switched between the server between the cluster system. Therefore, whether it is the hardware being executed, the user can see the virtual server. The IP address is actually migrating from a system to another. For external world, it can maintain a constant virtual server. So if the server has problems, an application that uses a specific IP address can continue to access the IP address, even if the IP address is now representing a different server. The virtual server allows users to find erroneous transfer operations, so users can continue to work without knowing what happened behind the scenes. Cluster components establish a series of components that require several components: cluster management software, server interconnect system, and shared disk systems. These components must be set, and combined with the cluster perception application to establish a cluster. In this section, you will learn about these different components and learn how they work together to build a cluster. In this chapter, you will learn how to set the SQL Server cluster. MSCS Cluster Management Software Cluster Management Software is actually a group of software tools that can be maintained, set, and start clusters. It consists of the next sub-component, which work together to maintain the functionality of the cluster, and perform erroneous transfer work: Node Manager maintenance of member relationships between cluster systems, transfer HeartBeats information to other Cluster member (node) (heartbeat is a message I also sent to I am Alive). If a node is stopped, another node will think it is no longer available, and will start to take over its work .Node Manager is one of the most critical parts of the cluster because it monitors the status of the cluster and the members of the cluster, and determines what action should be taken. Configuration Database Manager maintains the database settings of the cluster. This database retains all components in the cluster Tracking information, including abstract logical elements, such as virtual servoars, and entity elements (eg, shared disks). The database is similar to the login database of Windows NT / 2000. Resource Manager / Failover Manager launched Stop Mscs.manager / Failover Manager receives items from the Resource Monitor and Node Manager, such as lost nodes, new nodes, and more. Event processor is responsible for initializing the cluster, and routing event information between cluster components. Event Processor is also responsible By indicating that the Node Manager's new node is initialized as the extension of the cluster system. Communications Manager management of communication between the various nodes in the cluster. All nodes in the cluster must constantly communicate with other node to maintain the operation function of the cluster. If each node loses contact, the cluster status information will be lost, and the cluster will not work properly. Global Update Manager notifies the cluster status information to all nodes in the cluster, including the new and deletion of the cluster node, etc. Wait. Resource Monitor monitors the status of various resources in the cluster, and provides statistics. This information is used to determine whether there should be any error-mounted action in the cluster. Time service ensures that all nodes in the cluster can report the same system time .
If there is no Time Service, the order in which the event occurs will be mistaken, causing the system to make an incorrect decision. For example, if the node of the older version of the file is considered to be 2pm, and the node of the new version of the file thinks that it is 10 o'clock in the morning, the cluster will incorrectly think that the files in the first system are the latest. . The Server Interconnect server interconnect system refers to the connection between the collected nodes. Because the nodes in the cluster need to continuously communicate with components such as Time Service, Node Manager, etc., so maintaining high-speed connectors. The server interconnect system must be a reliable channel between the system to communicate. Servo interconnects are often executing TCP / IP or NetBIOS high-speed Ethaling. This setting is already sufficient, but you may want to use faster, high-speed interconnect systems than the Ethernet. Such interconnects can be bought from hardware vendors, some of which also provide such communication services like shared disks. Microsoft Web Site http://www.microsoft.com/HCL/ The compatible hardware list provides a complete list of general server interconnects. Another key component established by the shared disk system cluster is a shared-disk system. If multiple computer systems can access the same disk system, when the primary node is faulty, it can be transferred to other nodes. The shared disk system must allow multiple computer systems to access the same disk equivalent - in other words, each computer must be able to access all disks. In MSCS current versions, you can only have a system access disk each time, but the future version will allow multiple systems to access the disk simultaneously. There are already several common disk systems, and new disk technology is constantly developing. The SCSI disk subsystem has supported the function of multitiile initiators. With this feature, you can row with multiple SCSI controllers on the same SCSI bus, so that SCSI is quite suitable for clustering. In fact, SCSI is the first magnetic disc subsystem for cluster. Some new disk technology has recently been designed to support clusters such as Fiber Channel and some specialized solutions. Fiber Channel systems allow your computer to connect to a disk across a long distance. Most Fiber Channel systems support use multiple controllers on the same fiber circuit. Some RAID control cards have been developed or improved to support the cluster. If you do not modify or change settings, most of the disk control cards will not support the cluster. If you are getting tap, the content of the control card is allowed to allow the write to the memory to be stored in the memory, which will become a big problem with the cluster, as shown in Figure 12-3. In this case, each node has its own quick break, and we call this situation as a disk sharing (in Front of Disk Sharing) because two quick access sharing the same disk. When the system is faulty, since each control card has a quick time, but it is a failed system, the information in the quick tap will be lost. Therefore, if we use the internal control card to use the internal control card in the cluster setting, they should be set to the only reading mode. (In some cases, this setting may reduce the performance of some systems.) Figure 12-3 Clearing cards before the disk share
Solutions for shared disk system issues involve RAID, and quickly taken on the disk system itself. In this setting, all nodes share this quick break, which is called BEHIND THE DISK Sharing, as shown in Figure 12-4. The RAID mechanism and disk sharing system will be accessed by all control cards in the system, and read more and writing is safe.
Figure 12-4 Control Cards after the disc share of the disc is a new SCSI and the Fiber Channel disc system allow the RAID control card to be encapsulated in the magnetic disc, not in the computer system. This system provides excellent performance efficiency and fault tolerance. In fact, many of this type of RAID system provides backup control cards and getting more. There are many of the new RAID systems that have been used. Here are some detailed examples. The I / O subsystem has previously mentioned that there are several different types of I / O subsystem support clusters. The following is the most important type: SCSI JBOD is a system with a plurality of triggers (controllers) on the SCSI bus bar with JBOD (Just A Bunch Of ". In this setting, the disk has an independent address that must be set to the disk array using Windows 2000 or independent addressing. This system is usually not recommended. Internal RAID has a RAID control card on each server. The disadvantage of this subsystem is to avoid the loss of information, and the control card must be set to deactivate. The external RAID RAID control card is shared by the disk system in the bushes. Capacity and RAID logic are encapsulated in the disks, making a simple host bus card (HBA) connected to the external control card. Next, only two of the two RAID solutions are described in two sections. SCSI JBOD is only used when the cluster is small, and the cost problem is the main consideration. The internal RAID internal RAID control card is designed by the hard body to process the RAID process and will be taken to the host system. When using internal RAID, the shared disk system is located after RAID, as shown in Figs. 21-5.
Figure 12-5 Interior RAID Control Card
Since it is not shared on the control card, it is not shared, so the information in the quick tap will not be used in the system failure. This is a big problem in the Relevance Database Management System (RDBMS). When SQL Server writes data to a disk, the information may still be tap (because its operation is to write the disk control card to get the disk), but in the transaction record file, the information is written. If there is a fault problem at this time, SQL Server will not reply to these data blocks when trying to reply, because SQL Server believes that these materials have been written to disk. In this set type of error event, the database will thus be damaged. Therefore, the vendor must ensure that the quick-gathering the RAID control card for the cluster has been closed (or at least written to the function of the quick get). If the snap function has been closed, SQL Server does not receive a signal that has been completed before the data has not been written to the disk.
Note SQL Server is actions that perform all write to disk without using a buffer with a quick-tap mode. No matter how many archive systems are available, SQL Server will use them. Just like most RDBMS products, SQL Server completely ignores the file system.
In some cases, the control card will be greatly provided to provide the performance. Especially when you are using RAID 10 or RAID 5 because these RAID hierarchies produce additional burden when writing operations. To use the control card to write quickly in the cluster settings, you must use the external RAID system, let the quick get can be shared, the information will not lose when the error is missed. External RAIDs In an external RAID system, RAID hardware is outside the host system, as shown in Figure 12-6. Each server will have an HBA, which is to send more I / O to the RAID system as much as possible. The RAID system determines where the data actually stores.
Figure 12-6 External RAID subsystem
The external RAID subsystem is sometimes referred to as the cabinet RAID or chassis RAID because the RAID is performed in the disk cabinet. There are many advantages in the external RAID subsystem. It is not only an ideal solution for MSCs, but also an important comprehensive solution. The cabinet RAID has the following advantages: Easy to connect to the internal RAID, you need to pick up many cables from the RAID control card - each disk cabinet requires one. If you are using an external RAID, you only need a cable to connect the HBA with the external RAID control card, and the disk cabinet can be connected in the form of a daisy chain, and the control card is connected to this chrysanthemum ring, as shown in Figure 12 -7 shown. Even hundreds of disks that need to be connected, the external RAID can still simply complete this work. Allowing the cluster to use the external RAID to make it easier to set up a RAID solution. If you use an external RAID, you can have a quick getting and fault tolerance, but you don't have to worry about the quick consistency between multiple control cards (because there is only one quick time and a control card). In fact, if you are using an external RAID, you will not be safe. Although there is still risks when you get RDBMS data, this risk has been greatly reduced in the case of using an external RAID control card. To determine your external RAID system manufacturer support fast mirroring, when the memory wafer is wrong, the mirror can provide fault tolerance. Supporting more disks in large systems or high performance systems, sometimes a large number of disks must be planned. In Chapter 6, the necessity of the large number of disks is elaborated and learn how to plan the system's size; when learning the general implementation of the problem, it will also see it again in Chapter 36. The external RAID device allows you to connect hundreds of drives to a separate HBA. Internal RAID system, each control card can only connect a limited number of magneto, just like SCSI. Figure 12-7 Internal RAID Wiring VS. External RAID Wiring
In today, these can be used to support the coefficient of disk subsystem, the external RAID cabinet is most suitable for large clusters. Of course, the cost is also required, and some of the candies are too small to use external RAID. However, in the long-term implementation program, external RAID solutions can provide optimal performance, reliability, and management of your cluster. The type of application of the cluster application is executed on the system of MSCS, which can be divided into four categories: Cluster-UnaWare Applications This type of application does not generate any interaction with MSCs. Although they can be implemented in normal conditions, they may be turned off when they occur, and they cannot force the error to move to other nodes to continue. Cluster-aware Applications This type of application will realize the existence of MSCs. They can utilize MSCs to enhance their efficiency and admissibility. They have good response to the cluster event and only have few actions (even completely unnecessary) after the component fault and the error transfer occurs. SQL Server 2000 is a good example of a clustered perceived application. The cluster management application This type of application is used to monitor and manage the MSCS environment. Custom Resource Types This type of application provides self-reservations for other applications, services and devices. MSCS mode You can perform SQL Server 2000 Cluster Support and MSCs in different modes. In active / passive / passive mode, a server keeps the standby mode, ready to take over the main server when the system fails. In active / active / active mode, each server performs a different SQL Server library, no matter which server has an error, another server will take over it. In this case, a server will perform two databases. In this section, we will introduce the advantages and disadvantages of each model. Figure 12-8 Application type and MSCS
Active / Passive Clusters Active / Passive Cluster uses the primary node to perform the SQL Server application, and will use the server for the secondary node as a backup or standby server, as shown in Figure 12-9.
Figure 12-9 Active / passive cluster
In this setting, there is a server that is actually not used. This server may have been taken for any work for several months. In fact, most of the case is that the backup server never used. Because the secondary server is almost all move, it looks like an idle expensive device. Since the server cannot be used to perform other functions, in order to serve the user, you may need to purchase other devices, which makes the active / passive mode to spend money than what you think. Although the cost of active / passive mode is high, it also has advantages. In active / passive settings, if the primary node is wrong, all resources of the secondary node can be used to take over the primary node activity. If you are executing with a key task, the system requires a load specified traffic or a response time, which is quite important. In this case, the active / passive mode for you may be the correct choice. It is highly recommended that the secondary node has a completely identical hard body (that is, the number of RAMs is the same, the type of CPU is the same as the number). If the two nodes have exactly the same hardware, you can determine the performance efficiency of the secondary system can be almost the same as the main system. Otherwise, there may be problems with performance loss in the wrong transfer event. The active / active cluster is in the active / active cluster, each server performs an application, and also as a secondary server of another node, as shown in Figure 12-10.
Figure 12-10 Active / Active Clusters In this setting, each servo is both a primary node of some applications, as a secondary node of others. This is the most cost-effective setting, because there is no device being idle, waiting for other systems. Both systems can serve users. In addition, a separate passive node can be utilized as the secondary node of several major nodes. One disadvantage of active / active setting is that after the fault event occurs, the burden of the secondary node is aggravated, so that the results of surviving have a significant decline. Survive nodes now not only perform the original application, but also the application from the main node. In some cases, if it is not possible to endure the loss of performance, you must use active / passive settings. Examples of the cluster system are in this section, let's take a look at the four-use system examples of the MSCS. These examples will help you decide which type of cluster is most in line with your needs and the environment. Example 1 - High availability system with static load balancing This system can provide high availability on multiple applications on the cluster. However, if there is only one node left on all lines, the performance will be greatly reduced. The hardware resource usage of this system is the highest because each node can be accessed. Figure 12-11 shows the setting of such clusters, which is an active / active cluster.
Figure 12-11 High availability cluster with static load balancing
Each node in the cluster will notify the network in the form of a virtual server. Each node will reserve additional processing power on the setting, so the application of other nodes can be performed when the error transfer occurs. In other words, the service providing the error node user end is not interrupted by the additional resource and additional processing power. Example 2 - Heat replacement system with maximum usability This system provides maximum availability and performance through all system resources. The disadvantage of this setting is a large number of investments in hardware resources - and most of them are still used. One of the nodes will be treated as a major node and should be required to pay for all clients, and another node is idle. The idle node is actually a spare system used as a thermal state, only if the error event occurs, it will be accessed. If the primary node is wrong, this hot replacement node can immediately take over all the operations and continue to serve the user-ended requirements. Figure 12-12 shows this setting. This setting is still flexible for some applications with the most critical task nature. If your company is selling at the Internet, then your web trading server may operate in this setting. Since your business must rely on the success or failure of the system, spend more money to purchase an idle system to prepare for from time to time, it is actually worth it.
Figure 12-12 Heat replacement system with maximum usability
Example 3 - Local Server Cluster MSCS is a fairly flexible cluster plan, which can be seen from the local server cluster set. In this system, only the application you selected allows errors to move. As shown in Figure 12-13, you can specify that some applications are still available when their nodes are wrong, and other applications are not.
Figure 12-13 Local Server Cluster
This setting method is quite ideal when you intend to maximize the usage rate of hardware resources, but you must provide limited error-transfer capacity for applications with key task properties. In addition, this setting provides erroneous transfer to those clusters perceived applications, while also supporting nonset-in-aware applications. Example 4 - Virtual Server Uncomfortable Virtual Server Our last system example is not true uncomfortable, but it still uses MSCS support for virtual servo. This setting is actually a method used to organize resources, publish resources, as shown in Figure 12-14. The virtual server function allows you to use meaningful and descriptative names when naming resources, rather than a general server name list. In addition, MSCs will automatically restart the app or resources after the server is resolved. This feature is particularly useful for applications that do not provide internal mechanisms to restart. This architecture can also be said to be the best preparatory action to the cluster. Once you define a virtual server on a single node, you can simply join a secondary node without the need for a disturber definition. Figure 12-14 Virtual Server Unmarkable
SQL Server Cluster Settings After you install and set MSCs, the next step is to set SQL Server to execute in the cluster. As before, SQL Server 2000 has the ability to perceived cluster, and can make good use of clusters. In this section, let's first look at how to plan the cluster system, and then review the steps that must be performed in the cluster set SQL Server.
Note All benefits to the MSCS, the application must have the ability to perceived cluster. It has been said that the cluster-sensing application must understand the cluster architecture and can be misused in the error event. Not all applications support the cluster and have the ability to "escape from fan" when encountering problems in the cluster.
Planning Your Setup Planning SQL Server Card Set is to determine the hardware type to use, and which work mode is taken to take. It can form a plurality of hardware architectures of the cluster system, and the operation of the cluster can be selected to actively / passive mode or active / active mode. The selected mode will determine the hardware type and quantity you need. The active / passive cluster setting should be composed of the exact same two systems, each with the ability to master the overall workload. Since the active / passive mode does not use the secondary system during normal operation, the error will not be used after the error occurs, and the efficacy of the virtual server will remain unchanged. When the primary system error is transferred to a completely identical secondary system, the user will not detect any performance changes. The active / active cluster setting should consist of two systems that each perform a particular workload. If an error occurs, the survivation system will take over the workload of the error system. In this case, a single system should perform two jobs, and all users will face the problem of reduced performance. If your plan is cautious, the performance of the system may maintain an acceptable minimum limit, but in fact this efficacy is not insurance. When planning active / active cluster settings, you should have psychological preparation of loss of performance, see if you have to reduce some services, or to warn users in error transfer events, system efficiency may decline. The next step of setting the SQL Server Cluster function is to check or change some SQL Server settings. We will check these settings in the next three bars. Setting the recovery time When adjusting the SQL Server performance, you may have set the configuration option restore interval to other values, not the preset value of 0. Change this setting to increase the interval of checkpoints to improve performance, but will also increase the restoration time. In the cluster system, the preset value 0 should not be changed, which indicates that the recovery interval will be automatically set by SQL Server. (Having a system that can be misused is the main reason for using MSCs, this should be more important than the ability to consider it.) This setting allows the database to have a checkpoint almost every minute, and the maximum restoration time is about one minute. Related Information For more information, please refer to the online series and index the restore interval option.
Note Checkpoint operations Let SQL Server write data to the disks into the disk. Any modified information is not written in a disk when the system error is not written, and SQL Server is started at startup, and unrecognized transactions are restored, and the unidentified data is restored.
Setting SQL Server Executes on the active / passive cluster to create an active / passive cluster setting, you may have to modify a set value in the SQL Server. If your secondary server is exactly the same as the main server, you don't need to modify it. However, if the resources of the secondary server are less than the primary server, you must set the SQL Server configuration option MIN Server Memory to 0. This setting indicates that the SQL Server is allocated by the resource available to the system.
Related information To find relevant information, please refer to the line series and index the "Server Memory Options".
Setting SQL Server In active / active cluster set, you must set the SQL Server configuration option MIN Server Memory to 0. If the configuration option is set to manual, SQL SEVER may not be able to assign a memory after the error transfer. Since Windows 2000 is a virtual memory system, it can allocate more memories than actual available. In fact, this is also a common problem because of the patch action. For example, if each SQL Server system assigns a 75% system memory, the merged SQL Server service after the error transfer occurs, the SQL Server service will take 150% of the available memory. If the dynamic allocation is not possible, the system may be stagnant. status. Installing SQL Server in the cluster, installing SQL Server in the cluster, is quite similar to the SQL Server installer described in Chapter 7. Before starting this installation process, you must decide the location of SQL Server to install. You should install the SQL Server file on a shared disk controlled by the primary server. SQL Server's installation path and the Master library installation path should also set to the shared disk. You should also specify the network communication agreement that the cluster will be executed. Next, you can follow the steps to install SQL Server Error Transfer Units: Place the SQL Server 2000 CD into your disc player. If your system has automatic playback, SQL Server 2000's primary setting dialog will appear, as shown in Figure 12-15. Otherwise, you will need to manually execute Autorun.exe (found in the root directory of the CD).
Figure 12-15 SQL Server Settings Dialog
If you have not installed a must-have you have a must-have version, you have not already installed Microsoft Internet Explorer, or just want to check the necessary installation, press SQL Server 2000 necessary to install, SQL Server 2000 is necessary Install the dialog box. Press on the most suitable system to view its needs, and press to enter the installation on the option you need. If you already load the most appropriate software, you jump directly to step 3.
Note You must have MSCS before you start installing SQL Server.
Press SQL Server 2000 components to welcome the SQL Server 2000 installation wizard. If you are doing other Windows programs, close them. Press the next step to continue the installation process. On the computer name screen, press the virtual server and enter the server name, as shown in Figure 12-16. Press Next.
Figure 12-16 "Computer Name" screen
The user information picture appears. Enter your name and company name. Press Next. Software authorization consent. Press to accept its authorization consent and continue the installation process. In the installation screen, enter a product identification code of 25 flms (yellow label attached to the disc cartridge). Press Next. Error transfer cluster appears, as shown in Figure 12-17. Enter the IP address of the virtual server and press now. MSCS support subnet address. Press Next. In the cluster management screen, see the cluster definitions provided by SQL Server 2000. In the case of the preset, the computer is set to the priority node. Other available nodes appear in the Additional Node block. Press Next. In the remote information screen, enter the login credentials for the remote cluster node. This login credentials must have system administrator privileges for the remote node of the cluster. Press Next. In the execution of the Individual Name screen, select a preset execution individual or specify a naming execution individual. To specify the named execution individual, clear the preset value to verify the square, then enter the name of the execution individual. Press Next. Note The execution individual cannot be named DEFAULT or MSSQLServer or other names as SQL Server preserved keywords.
Figure 12-17 "Error Transfer Crowd" screen
In the installation type screen, select the type you want to install. The installation program will automatically set the preset value to the first available cluster disk resources you previously selected. However, if you need to specify a different cluster disk resource, press the Browse button under the information file and specify a path on different shared disk resources. Press Next. Then the verification mode dialog is shown, as shown in Figure 12-18. The setting in this picture will determine the security level of your SQL Server installation. You can choose Windows authentication mode or match mode (Windows Account Verification and SQL Server Account Verification). If you select a Windows authentication mode, the data library user's permissions will inherit from Windows user security settings. If you select a mode (Windows Account Verification, SQL Server Account Verification), you can individually define the security of the managed library user. If you select a matching mode (Windows Account Verification and SQL Server Account Verification), you must enter and confirm the SA (ie SQL Server System Administrator) login password. You can also leave a blank in the SA password field, but this will make your SQL Server security decrease.
Figure 12-18 "Verify Mode" screen
In the Start Copy Archive screen, press Next. Authorized mode screen appears. You have two SQL Server user-end authorization modes option, one is per server authorization, the other is to authorize each base. In each servo authorization mode, you must specify individual client access authorization (CAL) for a particular server, each authorization to allow a connection to the server. Regardless at any time, the maximum number of client computers that can be connected to the server is equal to the number of clients that you assign the server to the server. If you choose each server authorization, you must specify the number of connected client access to the server to the server. In each pedestal authorization mode, each computer that will access the SQL Server server requires a client access authorization. Once the computer is authorized, it can access any machine on the Internet without additional payment. If it is not sure which authorization mode is selected, press each serve. This authorization consent letter allows you to have an opportunity, one way to change settings to each base authorization mode. Press and continue to install the SQL Server application and data files. The SQL Server installation program will install the necessary files to your system and set the necessary components. The installation may take a few minutes, or longer, it depends on your system. When the installation completion screen appears, select the option to restart the computer and press it. As you can see, set the SQL Server cluster is quite easy. Once the cluster is set, no additional setting steps are needed. The user will access SQL Server through the IP address. This IP address will be redistributed to the secondary node after the error transfer. The remaining program should be considered, we will discuss in the section "Beyond MSCS> section later. Most applications using three-layer application are directly connected to the database. The application proposes a transaction, and the database responds to the transaction. In the system error event, the fair will also perform the status of failure. In most cases, this setting will not have any problem - if the transaction is not completed, you will want the application to fail. However, if you use the wrong transfer, the database will return to the available status in a short period of time and continue to respond to the transaction after the error. You can use a Three-Tier's application to ensure that this fast recovery service can be used. In a three-layer application, the interposer can aware of the server has stopped responding and waits for a while, then re-submits the transaction. Users may feel that the completion of the transaction seems to be a little delay, but the delay completion is much better than the failure of the transaction. To achieve this goal, the application must be noticed when the connection failed between the server and the application, and the application must be noticed and knows the connection. The application should also use a message dialog or otherwise notify the end user using a message dialog or other means. Using a three-layer application, it is no longer an unreachable ideal for building a sky. The application must have the ability to perceived cluster, and know that the virtual server will continue to work back to work. Together with the MSCS architecture, you can provide sturdy applications and data. Beyond MSCs We have already introduced the basic concepts of MSCs and how SQL Server works on this architecture. We have also seen how SQL Server is scared in some hardware and software errors and re-execute transactions in a short time. To achieve this fault, you need to use MSCS, but also take some measures. Two steps are quite important: First, the rules and index execution of backups, the other is to prepare a disaster reduction plan. The procedure for system backup and how to prepare a disaster reduction plan will be introduced in Chapter 32 and 33.
