ASM gives you performance and redundancy through striping and mirroring.
Examining ASM Striping
ASM systems store your database files on ASM disks. The manner in which you place your database files on ASM disks plays a critical role in the resulting performance. For optimal I/O performance, ASM stripes its files across every disk that is part of its disk group. This means that all the disks in a disk group must be of the same type and performance capacity.
ASM offers two types of striping, with the choice depending on the type of database file. Coarse striping uses a stripe size of 1MB, and you can use coarse striping for every file in your database, except for the control files, online redo log files, and flashback files. Fine striping uses a stripe size of 128KB. You can use fine striping for control files, online redo log files, and flashback files.
Examining ASM Mirroring
Disk mirroring gives us data redundancy. This means that, should you lose a disk, you can use the mirror disk to continue operations. This process is not like an OS-level mirroring scheme, but they both provide redundancy for your database. The difference is that OS-based LVMs mirror entire disks, whereas ASM mirrors extents. This means that when ASM allocates an extent (the primary extent, in contrast to a mirrored extent), it also allocates a mirror copy to one of the disks in the same disk group.
When a disk in a group fails, ASM rebuilds the failed disk using the mirrored extents from other disks in the group. When ASM reconstructs a failed disk, the storage system takes a small performance hit, because ASM requires some extra I/O to reconstruct the failed device.
Failure Groups
Disk failure is not the only way in which you can lose a disk. You can also lose a disk if shared resources, such as SCSI disk controllers, fail. When one of these fails, you cannot access any of the connected disks. A set of disks that fail because they all share a common resource, such as a disk controller, is a failure group. You ensure redundancy by mirroring disks on a separate failure group.
To avoid problems, ASM will not place a primary extent and its mirror copy in the same failure group. This means that even if a failure group loses several disks, ASM can survive the disaster and reconstruct the lost disks from the mirror copies that are in a different failure group.
Types of ASM Mirroring
ASM supports three forms of disk mirroring, each with a different level of data redundancy. External redundancy doesn’t have failure groups, and thus is effectively a no-mirroring strategy.
Normal redundancy provides two-way mirroring of all fields in a disk group.
High redundancy provides three-way mirroring, which results in three failure groups, with a disk controller for each.
Comments