Home > Workload Solutions > Oracle > White Papers > Oracle ASM on ScaleIO Best Practices > How many ASM diskgroups?
Oracle recommends just two diskgroups by default:
By default, redo logs and control files are mirrored to both the DATA and FRA diskgroups.
For DBAs planning to use ScaleIO snapshots, a two-diskgroup solution should be considered the minimum, but control files and redo logs should NOT be mirrored to both diskgroups.
When using ScaleIO snapshots, it can be beneficial to snapshot the DATA and FRA diskgroups independently. This allows for the FRA diskgroup to be restored at a time slightly ahead of DATA, allowing Oracle to perform additional database recovery through the application of archivelogs.
When using a mix of storage pools with both high capacity spinning disks and high performance solid state devices additional diskgroups should be considered, which correspond to the different performance storage pools.
It is important to ensure that control files, redo log files, temp and undo reside exclusively in a pool with high performance.
Since Oracle Database 11gR2 it is recommended to store the OCR and Voting Files in a standalone ASM diskgroup, as shown in Table 3.
Table 3. ASM diskgroups
ASM diskgroup |
Purpose |
Database tablespaces |
Storage pool characteristics |
RAM/Flash cache |
GRID |
Cluster Voting Disk, OCR and ASMSPFile |
N/A |
Performance |
No |
DATA |
Database data files, temp files, redo logs and undo |
SYSTEM, SYSAUX, DATA, INDEX, USERS, TEMP, UNDO |
Performance |
No |
FRA |
Archive logs and backups |
N/A |
Capacity |
No |