ManagingHighAvailability•Chapter6 197 Although there is no performance benefit when using this RAID level, the data is fully protected since if either hard drive fails the data is still available by the use of the hard drive that has not failed. This is one of the most expensive RAID options available to you because you are paying for two hard drives and only getting the space and speed of one hard drive. When the failed hard drive is replaced all the data is copied from the working hard drive to the new hard drive. RAID 5 is the most common RAID level in use. It provides the best of both RAID 0 and RAID 1, but at a lower cost per gigabyte. A RAID 5 array requires at least three hard drives. RAID 5 writes data to all the disks except for one. The last disk is used to store parity information for the data on the other disks in the array. This allows the system to continue to function in the event of a disk failure. It also allows the system to re-create the missing information once the failed hard drive is replaced. Because this parity data must be calculated for each write operation the writes to a RAID 5 array will be slower than some of the other RAID levels that are available. RAID 6 is a new RAID level that is really starting to show itself in the database world. A RAID 6 array requires at least four hard drives. RAID 6 is very similar to RAID 5, except that it uses two separate parity drives. This allows the RAID array to survive two different disk failures without running the risk of losing any data on the array. Like the RAID 5, array a RAID 6 array will be slower than some of the other RAID levels that are available because of this parity calculation. RAID 10 is the highest performing, highest redundancy, most expensive RAID solution that is available to you. A RAID 10 array requires at least four hard drives. A RAID 10 array is in essence two RAID 0 arrays for performance that are then mirrored together for redundancy. Because of the mirroring that is being done this is the most expensive RAID level available per gigabyte. Because of this cost many people will prefer to use RAID 5 or RAID 6. Ex a m Wa r n i n g Microsoft has started putting RAID questions on the exams. A basic understanding of the RAID levels is key to passing those questions. When dealing with the storage for many of the high-availability solutions, we will be looking at a high-end storage solutions such as a SAN. Some of these 198 Chapter6•ManagingHighAvailability solutions require these high-end storage solutions. When using these high-end solutions keep in mind that your hard drives could be sharing the physical spindles on the SAN with other systems in your environment. When you are troubleshooting systems using these SAN solutions, make sure that you know what else is using the hard drives, even if it not a SQL Server. If you have a file server on the same spindles as a SQL Server, that can be fine, unless the file server is defragging the hard drive, in which case the disks will perform poorly without no indication on the SQL Server as to the problem. When selecting the hardware for your backup server, you can usually cut a few corners to save a couple of bucks. The goal of your backup server is to run the system. It doesn’t have to run it at the same speed as your production environment does, it just needs to keep it running while the primary or active system is being repaired or rebooted. As an example, if your system runs fine on two processors but runs great on eight, get a two-processor server for the backup machine. This will help keep the costs of the backup server a bit more reasonable. The licensing costs are different for each technique that we will be discussing; therefore the license ramifications will be covered in each section throughout this chapter. SQL High-Availability Options There are many different high-availability options that are available to you as a production database administrator. Each of these various options has its strong points and its weak points, and in addition to knowing how to use each method, it is very important that you know when each method should be used. Log Shipping SQL log shipping is a tried-and-true method of keeping your SQL Server online. Microsoft has supported log shipping within SQL Server since at least SQL Server 2000. By using standard T/SQL backup and restore commands you could have set up log shipping back in the SQL 7 and prior days. Log shipping (also known as transaction log shipping) is the process by which you take a full backup of your database and restore it to the destination machine. You then back up the transaction log every few minutes (usually between 5 and 15 minutes) and then copy the log backup to the remote machine. After the file has been copied you then restore the transaction log to the remote machine or machines, rolling forward any transactions that are in the transaction log. This gets you an exact duplicate of the primary database. ManagingHighAvailability•Chapter6 199 Log shipping does not require special hardware, nor does it require that the backup server be configured like the production server. However it does make it much easier when the two servers are using like hardware, and most specifically when the hard drives are laid out exactly the same. SQL Server 2008 includes log shipping in all the Editions of SQL Server that you have to pay for—in other words, Web, Workgroup, Standard, and Enterprise Editions. This is a big change from SQL Server 2000 when log shipping was an Enterprise-only feature. TE s T Da y Tip Unless a question specifically mentions the Edition of SQL Server that you are using, licensing or features that are missing between editions are not going to be the answer. Unless stated otherwise you can usually assume that you have the Enterprise Edition of all products involved in the question. To begin log shipping, connect to the server you wish to be the source of the log shipping in the Object Explorer. Right-click the database to bring up the context menu and select Tasks, then Ship Transactions Logs as shown in Figure 6.1. You can also get to this Transaction Log Shipping menu by selecting Properties from the first context menu. 200 Chapter6•ManagingHighAvailability On the screen that appears, check the check box at the top of the window that says Enable this as a primary database in a log shipping configuration. This will enable the Backup Settings button. Click this button and specify the two paths listed, as well as file deletion settings. These folders must exist before you can click OK on the main settings page. Keep in mind that SQL will delete these log files even if they have not been restored to the backup server so don’t set them too short. The backup job name will be the name of the SQL Agent job with which the backups are taken. In the example shown in Figure 6.2 the backups will be made to the C drive of the SQL1 server, which is the primary server. The backups will be kept for 72 hours with an alarm raised if no backup is taken within an hour. You can enable backup compression on the transaction log backups; however this usually is not needed unless your backup server is across a WAN. TE s T Da y Tip The exams are designed to make sure that you know the official way to perform tasks, which is not always the quickest or easiest way. This is a key distinction that you should be aware of when you are taking the exams. Figure 6.1 Database Context Menu ManagingHighAvailability•Chapter6 201 After you complete the backup settings, you can tell the database server to which Instances you want to ship the logs. Single-click the Add button in the middle of the window, then click Connect on the new window. On the first of three tabs on this window you can tell SQL Server if you want SQL to handle the full backup automatically, use an existing full backup, or whether you have manu- ally restored the full backup. On the second tab you give SQL Server the folder on the backup server to which you wish to have it copy the files. You also give it the name of the SQL Agent job that will copy the files from the primary system to the backup system. On the third tab you tell SQL Server which state to leave the database in—No Recovery or Standby. You can also tell it how long to wait before restoring a log file, and when to raise an alert if no log has been restored. You also name the restore job that the SQL Server Agent will run. After clicking OK you are taken back to the Log Shipping page of the database properties as shown in Figure 6.3. Figure 6.2 Transaction Log Backup Settings . SQL will delete these log files even if they have not been restored to the backup server so don’t set them too short. The backup job name will be the name of the SQL Agent job with which the. manu- ally restored the full backup. On the second tab you give SQL Server the folder on the backup server to which you wish to have it copy the files. You also give it the name of the SQL Agent job. with which the backups are taken. In the example shown in Figure 6.2 the backups will be made to the C drive of the SQL1 server, which is the primary server. The backups will be kept for 72 hours