Figure 2-2 represents another mirror configuration. In this case, the master device, user databases, and transaction log are all stored on different partitions of the same physical device and are all mirrored to a second physical device.
The configuration in Figure 2-2 provides nonstop recovery from hardware failure. Working copies of the master and user databases and log on the primary disk are all mirrored, and failure of either disk does not interrupt Adaptive Server users.
Figure 2-2: Disk mirroring for rapid recovery
With this configuration, all data is written twice, once to the primary disk and once to the mirror. Applications that involve many writes may be slower with disk mirroring than without mirroring.
Figure 2-3 illustrates another configuration with a high level of redundancy. In this configuration, all three database devices are mirrored, but the configuration uses four disks instead of two. This configuration speeds performance during write transactions because the database transaction log is stored on a different device from the user databases, and the system can access both with less disk head travel.
Figure 2-3: Disk mirroring: keeping transaction logs on a separate disk
