ASM Failure group is defined as protection against the problems in disk in ASM disk groups. Oracle provides disk redundancy with failure groups.
If the failure group is not defined in non-Exadata environments, the failure group for each disk is itself. In this case, the failure group of a block on a disk is copied to a different disk.
In the Exadata environment, the failure group for disks on each cell node is cellnode itself. In this way, the blocks on the disk in the cell node is also present on another cellnode, so even if the cellnode is turned off, there will be no problem.
We can define failure groups when creating a disk group. First you should understand disk redundancy types.
Disk Redundancy Types
High Redundancy : data is stored in 3 copies, including itself. A disk group requires at least 3 failure groups. Each copy must be on different failure group. In this way, even if the two failure groups fails, there will be no data loss and database continues to work.
Normal Redundancy : data is stored in 2 copies, including itself. A disk group requires at least 2 failure groups. Each copy must be on different failure groups.
External Redundancy: The copy of the data is not stored. You can choose this type if there is already a RAID structure on the storage.
Lets create a disk group with 3 failure groups and normal redundancy as follows.
|
1 2 3 4 5 6 7 8 9 10 11 |
CREATE DISKGROUP DATA NORMAL REDUNDANCY FAILGROUP fg1 DISK '/dev/sda' NAME disk1, FAILGROUP fg2 DISK '/dev/sdb' NAME disk2, FAILGROUP fg3 DISK '/dev/sdc' NAME disk3 ATTRIBUTE 'au_size'='4M', 'compatible.asm' = '11.2', 'compatible.rdbms' = '11.2', 'compatible.advm' = '11.2'; |
According to the failure group definitions in this disk group, a copy of the blocks on /dev/sda will be on /dev/sdb or /dev/sdc. But the blocks of /dev/sda disks will never be on this disk. This is the case for normal redundancy.
The block distribution in this disk group will be as follows. Three failure groups and normal redundancy.
When a disk in this disk group becomes corrupted, the disk group will be available without data loss.
If no external redundancy is used in disk group, it is recommended that the disks in a disk group have the same size and have the same number of disks with failure groups. This is mandatory in version 12c. In the case of a different behavior, disk space problems may occur.
It is not possible to create a copy of all blocks between disks in different failure groups with different disks on different sizes.
