High Availability for Microsoft SQL Server Using Doble-Take 4.x High Availability for Microsoft SQL Server Using Doble-Take 4.x pblished April 2000 NSI and Doble-Take are registered trademarks of Network Specialists Inc. All other prodcts are trademarks of their respective companies. 1996 2000 NSI Software
Table of Contents Introdction... 1 Reqirements... 1 Monitoring Mltiple SQL Databases... 1 Application Note Otline... 2 Protecting Yor SQL Data... 3 Install Software on the Sorce...3 Install and Configre Software on the Target...4 Configre and Begin Mirroring and Replication...6 Configre Failover and Begin Failre Monitoring...9 Monitoring Failover... 13 When the Sorce is Online...13 When the Sorce Fails and Failover is Initiated...15 When the Sorce Fails and Failover is Completed...15 Restoring Yor SQL Data... 16 NSI Software Contact Information... 21 High Availability for Microsoft SQL Server Using Doble-Take 4.x
Introdction Microsoft SQL Server is a scalable, reliable, flexible, and high-performance relational database management system for Microsoft Windows 2000/NT server-based systems. NSI Software s Doble-Take provides real-time enterprise data protection and replication. Doble-Take can be sed to provide high availability for yor SQL server. This docment describes the steps necessary to configre Doble-Take to provide high availability for Windows 2000/NT servers rnning Microsoft SQL Server versions 6.5 or 7.0. These procedres allow a secondary server to assme the identity and role of a failed SQL server while maintaining the availability of SQL services with minimal disrption or data loss. To complete these instrctions, yo will install Microsoft SQL Server and Doble-Take, and configre Doble-Take for replication and failover. De to the complexities of these applications, this docment is intended for network administrators with experience installing, configring, and maintaining network applications inclding Doble-Take and Microsoft SQL Server. Reqirements Two servers that meet one of the following operating system reqirements: Microsoft Windows NT 4.0 with Service Pack 4 or higher Microsoft Windows 2000 NOTE: The two servers shold both be rnning the same operating system. Althogh cross-platform mirroring and replication are available, NSI Software recommends that the two servers be the same platform for effective failover and failback. It is recommended that both sorce and target servers be standalone servers. (Yo may experience problems with promotion and demotion dring failover if either of the machines are Primary or Backp Domain Controllers.) Two licensed copies of Microsoft SQL Server 6.5 or 7.0 Two licensed copies of Doble-Take 4.x Both servers mst be connected to the same physical network Monitoring Mltiple SQL Databases Doble-Take allows yo to configre one target to monitor and failover for one or more sorce machines. In a one-to-one configration, yo will want to replicate yor SQL data to the same location on the target so that failover is atomatic. In a many-to-one configration, each SQL data store will need to be replicated to a niqe location and then renamed to the corresponding SQL directory on the sorce before failover occrs. This application note focses on a single SQL server being replicated to a single target. High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 1 of 21
Application Note Otline Protecting Yor SQL Data 1. Install Software on the Sorce a. Install Microsoft SQL version 6.5 or 7.0. b. Install Doble-Take 4.x 2. Install and Configre Software on the Target a. Install Doble-Take 4.x. b. Configre Doble-Take. c. Install Microsoft SQL 6.5 or 7.0. d. Set SQL services to manal startp. 3. Configre and Begin Mirroring and Replication a. Identify the SQL data to protect and create the Doble-Take replication set. b. Connect the replication set to the target and begin mirroring and replication. 4. Configre Failover and Begin Failre Monitoring a. Create scripts to be sed dring the failover and failback processes. b. Configre failover. c. Begin failre monitoring. Monitoring Failover 1. When the Sorce is Online 2. When the Sorce Fails and Failover is Initated 3. When the Sorce Fails and Failover is Completed Restoring Yor SQL Data 1. Rebild the sorce machine, if necessary. 2. Stop the SQL services on the target. 3. Initiate failback. 4. Bring the sorce online. 5. Restore the newer data from the target to the sorce. 6. Start the SQL services on the sorce. High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 2 of 21
Protecting Yor SQL Data Install Software on the Sorce 1. Install Microsoft SQL Server version 6.5 or 7.0 on the sorce, if it is not already installed. 2. Record the drive and directory where Microsoft SQL is installed. For example, the defalt directories for SQL 6.5 and 7.0 are <drive>:\mssql and <drive>:\mssql7, respectively. SQL Installation Drive and Directory: 3. Install Doble-Take 4.x on the sorce machine sing the installation defalts. NOTE: Verify that the transacational applications option is selected on the Doble-Take Optimizations screen since Microsoft SQL Server is a transactional database application. See the Doble-Take gide Getting Started for frther details. High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 3 of 21
Install and Configre Software on the Target 1. Install Doble-Take 4.x on the target sing the installation defalts. NOTE: Verify that the transacational applications option is selected on the Doble-Take Optimizations screen since Microsoft SQL Server is a transactional database application. See the Doble-Take gide Getting Started for frther details. 2. In Control Panel, Services, doble-click the Doble-Take service. 3. Mark the check box Allow Service to Interact with Desktop and click OK. 4. Install Microsoft SQL version 6.5 or 7.0 on the target sing the same drive and directory specifications recorded in step 2 of the previos section. High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 4 of 21
5. Using the table below, identify the services that yo will need to set to manal startp so that all SQL files are closed on the target and the Doble-Take sorce can replicate the changes. SQL 6.5 SQL 7.0 Windows 2000 Windows NT 4.0 Distribted Transaction Coordinator Message Qeing MSSQLServer SQLExective MSDTC Message Qeing MSSQLServer SQLExective Distribted Transaction Coordinator Message Qeing MSSQLServer SQLServerAgent MSDTC Message Qeing MSSQLServer SQLServerAgent NOTE: If a failre shold occr, the failover and failback scripts that yo will be creating will control the stopping and starting of the SQL services. High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 5 of 21
Configre and Begin Mirroring and Replication 1. Select Start, Programs, Doble-Take, Management Console. 2. Doble-click yor sorce machine to log on. 3. Right-click the sorce machine and select Properties. 4. Select the Startp tab. 5. By defalt, Perform Remirror After Ato-Reconnect will be selected. Clear this option so that the sorce does not remirror files after failback. If this option is selected, the sorce cold remirror files after a failback and potentially overwrite newer files on the target machine. 6. Click OK. High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 6 of 21
7. Right-click yor sorce machine and select New, Replication Set and enter the desired name for the replication set. 8. Select the SQL data yo wish to protect. Most likely, this will only inclde the SQL data and log files. It is not necessary to replicate the application files since they are already exist on the target machine. Select the following directories: <drive>:\<sql Install Directory>\LOG <drive>:\<sql Install Directory>\Data Select any other directories (even if on different drives) that yo may have created to store SQL data files. 9. Right-click the replication set name and select Save to save the replication set. High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 7 of 21
10. Drag and drop the replication set onto the target. The Connection Manager dialog box opens. 11. The Sorce Server, Target Server, Replication Set, and Rote fields will atomatically be poplated. If yo have mltiple IP addresses on yor target, verify the Rote field is set to the correct network path. (For detailed information on connecting a sorce and target, reference Doble-Take s User s Gide.) 12. Select the One To One mapping so that the replication set data is transmitted to the same directory strctre on the target. 13. Click Connect to start the mirror and replication processes. High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 8 of 21
Configre Failover and Begin Failre Monitoring 1. If a failre occrs, yo will want to have the SQL services start on the target machine atomatically. To do this, create a batch file called postover.bat sing the sample batch file below. Save the batch file to the same directory where yor Doble-Take files are installed. POSTOVER.BAT rem This file is configred by defalt to rn with SQL 7.0 on Windows 2000 rem Yo may need to comment the defalt lines ot and ncomment the set of rem lines for the SQL and Windows version yo are sing rem If yo are sing SQL 6.5 and Windows 2000 se the following for commands. rem net start Distribted Transaction Coordinator rem net start Message Qeing rem net start MSSQLServer rem net start SQLExective rem If yo are sing SQL 6.5 and Windows NT 4.0 se the following for commands. rem net start MSDTC rem net start Message Qeing rem net start MSSQLServer rem net start SQLExective rem If yo are sing SQL 7.0 and Windows 2000 se the following for commands. net start Distribted Transaction Coordinator net start Message Qeing net start MSSQLServer net start SQLServerAgent rem If yo are sing SQL 7.0 and Windows NT 4.0 se the following for commands. rem net start MSDTC rem net start Message Qeing rem net start MSSQLServer rem net start SQLServerAgent High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 9 of 21
2. After a failre is resolved, yo will be ready to bring yor sorce back online. At this time, yo will want to stop the SQL services on the target atomatically. To do this, create a batch file called preback.bat sing the sample batch file below. Save the batch file to the same directory where yor Doble-Take files are installed. PREBACK.BAT rem This file is configred by defalt to rn with SQL 7.0 on Windows 2000 rem Yo may need to comment the defalt lines ot and ncomment the set of rem lines for the SQL and Windows version yo are sing rem If yo are sing SQL 6.5 and Windows 2000 se the following for commands. rem net stop Distribted Transaction Coordinator rem net stop Message Qeing rem net stop MSSQLServer rem net stop SQLExective rem If yo are sing SQL 6.5 and Windows NT 4.0 se the following for commands. rem net stop MSDTC rem net stop Message Qeing rem net stop MSSQLServer rem net stop SQLExective rem If yo are sing SQL 7.0 and Windows 2000 se the following for commands. net stop Distribted Transaction Coordinator net stop Message Qeing net stop MSSQLServer net stop SQLServerAgent rem If yo are sing SQL 7.0 and Windows NT 4.0 se the following for commands. rem net stop MSDTC rem net stop Message Qeing rem net stop MSSQLServer rem net stop SQLServerAgent NOTE: These sample batch files are available on the NSI Software web site at www.nsisoftware.com/download/sqlscrp.exe. 3. Select Start, Programs, Doble-Take, Failover Control Center. High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 10 of 21
4. Select the target machine from the list of available machines. If the target yo need is not displayed, click Add Target, enter the machine name, and click OK. Sorce Name: Larry_Legend IP Address: 206.31.4.55 Target Name: Perses IP Address: 206.31.4.54 Configration Before Failover 5. To add a monitor for the selected target, click Add Monitor. Type the name of the sorce machine and click OK. The Monitor Settings window will open. 6. In the Monitor Settings window, mark the IP address that is going to failover and verify that Adding Sorce Identity to Target is selected. High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 11 of 21
7. Click Scripts and specify the scripts that were created in steps 1 and 2 on page 9 and page 10. 8. Click OK to go back to the Monitor Settings dialog box. 9. Click OK to begin monitoring the sorce machine. In the event of a sorce machine failre, yor target machine is now ready to stand in for yor sorce. Failed Sorce Target Name: Larry_Legend and Perses IP Address: 206.31.4.55 and 206.31.4.54 Configration After Failover High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 12 of 21
Monitoring Failover Now that replication and failover monitoring are configred and started, yo will need to know if and when there is a problem. Since it can be essential to qickly know the stats of yor machines, Doble-Take offers varios methods for monitoring the stats of failover. When the Failover Control Center is rnning, yo will see for visal indicators: The Failover Control Center Time to Fail contdown The Failover Control Center stats bar located at the bottom of the window The Failover Control Center colored bllets to the left of each IP address and sorce machine The Windows desktop icon tray containing a failover icon NOTE: Yo can minimize the Failover Control Center and, althogh it will not appear in yor Windows taskbar, it will still be active and the failover icon will still appear in the desktop icon tray. When the Sorce is Online Dring normal operation when the sorce machine is online and responding, the failover indicators identify that the monitored sorce is still commnicating with the target. The Failover Control Center Time to Fail conter is conting down and resetting each time a heartbeat is received from the sorce machine. The Failover Control Center stats bar at the bottom of the window indicates that the target machine is monitoring. The Failover Control Center bllet before each monitored IP address and sorce machine name is green. The Windows desktop icon tray contains a failover icon depicting two compters; one red and one green. This icon indicates that both the sorce and target machines are online and active. High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 13 of 21
When the Time to Fail vale has decreased by 25% of the entire timeot period, the bllet changes from green to yellow, indicating that the target has not received a response from the sorce. The yellow bllet is a cation signal. If a response from the sorce is received, the contdown resets and the bllets change back to green. If the contown reaches zero withot the target receiving a response from the sorce, failover begins. High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 14 of 21
When the Sorce Fails and Failover is Initiated If the sorce machine experiences a failre, failover is initiated. The failover indicators identify that a monitored sorce is in the process of being failed over to its target. The Failover Control Center Time to Fail contdown vale is 0. The Failover Control Center stats bar at the bottom of the window displays the sorce machine and IP address crrently being assmed by the target. NOTE: If the failover process completes qickly, this indicator may appear and disappear before yo have time to read it or actally see it. The Failover Control Center bllet before each failed IP address and machine name is yellow. The Windows desktop icon tray contains a failover icon depicting two compters; one red and one green. When the Sorce Fails and Failover is Completed When the failover process is complete, the visal indicators are again pdated. The Failover Control Center Time to Fail contdown is replaced with the Failed Over message. The Failover Control Center stats bar at the bottom of the window indicates that monitoring has contined. The Failover Control Center bllet before each failed IP address and machine name is red. The Windows desktop icon tray contains a failover icon with one red compter. This icon indicates that at least one monitored sorce is in a failed over state. Once failover is complete, any clients logging into the SQL server will be atomatically directed to the target. Depending on the type of client software being sed, SQL clients with an exsiting connection may have to be restarted in order to establish a connection with the target. NOTE: For additional detailed information on failover and other monitoring tools, see Doble-Take s User s Gide. High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 15 of 21
Restoring Yor SQL Data If yor sorce experiences a failre, sch as a power, network, or disk failre, yor target machine will stand in for the sorce while yo resolve the sorce machine isses. Dring the sorce machine downtime, data is pdated on the target machine. When yor sorce machine is ready to come back online, the data is no longer crrent and mst be pdated with the new data on the target machine. 1. Verify that yor sorce machine is not connected to the network. If it is, disconnect it. 2. Resolve the sorce machine problem that cased the failre. NOTE: If yo mst rebild yor hard drive, contine with step 3. If yo do not need to rebild yor hard drive, contine with step 7 below. 3. Install Windows 2000/NT. Since yor sorce machine is not connected to the network, go ahead and se the sorce s original name and IP address. 4. Install Doble-Take 4.x sing the installation defalts. NOTE: Verify that the transacational applications option is selected on the Doble-Take Optimizations screen since Microsoft SQL Server is a transactional database application. See the Doble-Take gide Getting Started for frther details. 5. Install SQL sing the same drive and directory settings recorded in step 2 of the first section. High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 16 of 21
6. Using the table below, identify the services that yo will need to set to manal startp so that all SQL files are closed on the target and the Doble-Take sorce can replicate the changes. SQL 6.5 SQL 7.0 Windows 2000 Windows NT 4.0 Distribted Transaction Coordinator Message Qeing MSSQLServer SQLExective MSDTC Message Qeing MSSQLServer SQLExective Distribted Transaction Coordinator Message Qeing MSSQLServer SQLServerAgent MSDTC Message Qeing MSSQLServer SQLServerAgent 7. Verify that SQL is not rnning on the sorce. The SQL services mst not be rnning at this time. Depending on the type of failre, yor services may be set to manal startp bt cold still be rnning. Stop yor SQL services and set them to manal. 8. Select Start, Programs, Doble-Take, Failover Control Center. 9. Select the target machine that is crrently standing in for the failed sorce. High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 17 of 21
10. Select the failed sorce and click Failback. The pre-failback script entered dring the failover configration stops the SQL services on the target so that no additional changes can be made. 11. Yo will be prompted to determine if yo want to contine monitoring the sorce server. Do not choose Contine or Stop at this time. 12. Connect the sorce machine to the network. 13. Open the Doble-Take Management Console (Start, Programs, Doble-Take, Management Console), highlight the sorce machine, and ensre that the SQL replication set is not connected so that replication does not contine after failback is complete. If the replication set is connected, disconnect it by right-clicking it and selecting Disconnect. 14. After the sorce is back online, select whether or not yo want to contine monitoring this sorce machine. High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 18 of 21
15. To begin the restoration process, open the Doble-Take Management Console and select Tools, Restoration Manager. NOTE: Yo can also rn the Doble-Take DTCL atomated restoration script, which can be fond in Doble-Take s User s Gide, to complete the remaining steps in this section. High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 19 of 21
16. Complete the appropriate fields as described below. Original Sorce The name of the sorce machine where the data original resided. Restore From The name of the target machine that contains the replicated data. Replication Set The name of the replication set to be restored. Restore To The name of the machine where yo the data will be restored. This may or may not be the same as the original sorce machine. 17. Identify the correct drive mappings for the data and any other restoration options necessary. For detailed information on the restoration options, see Doble-Take s User s Gide. 18. Verify that the selections yo have made are correct and click Restore. The restoration procedre time will vary depending on the amont of data that yo have to restore. 19. After the restoration is complete, start the SQL services on the sorce machine. 20. Reestablish the Doble-Take SQL replication set connection. At this time, yor data is restored back to yor sorce machine, the sorce machine is again the primary SQL server, and, if yo selected to contine failover monitoring, the target is available to stand in for the sorce in the event of a failre. High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 20 of 21
NSI Software Contact Information Web Site Company and prodct information, inclding a list of recent press releases, is available on or web site at www.nsisoftware.com. Crrent pdates of all prodcts can be fond at www.nsisoftware.com/pdates. Technical Spport When contacting s for technical spport, please inclde the following identifying information, in addition to the description of yor problem or sitation yor name, yor company name, and a phone nmber where yo can be reached; yor NSI prodct name, version nmber, and its serial nmber and activation code; and a brief description of yor operating system and network configration in which the NSI prodct is operating. Web Site Spport An online spport form is available at www.nsisoftware.com/pages/spportform.htm. E-mail Spport Or e-mail address is spport@nsisoftware.com. Phone Spport Technical spport hors are Monday throgh Friday, 8:00 a.m. - 6:00 p.m. (EST). Cstomers within the U.S., please call (800) 775-8674. Cstomers otside of the U.S., please call +1 317 598-2066. Twenty-for-hor emergency spport is also available to all NSI cstomers. Unless yor service agreement covers 24-hor spport, there is a charge for each instance. After a message has been left, a technician will be paged and will retrn yor call. Sales and Corporate Office For information abot NSI Software or or prodcts, please contact: (800) 775-4674, if calling from within the U.S. +1 201 656-2121, if calling from otside the U.S. Bsiness hors are Monday throgh Friday, 9:00 a.m. - 5:00 p.m. (ET). Please address all other correspondence as follows: NSI Software 80 River Street, Site 5B Hoboken, NJ 07030 Direct fax correspondence to (201) 656-2727. Docmentation For comments or sggestions abot NSI Software docmentation, e-mail the docmentation manager at docmentation@nsisoftware.com. High Availability for Microsoft SQL Server Using Doble-Take 4.x Page 21 of 21