Sizing Exadata - Oracle Exadata Recipes: A Problem-Solution Approach - page 116

Table 5-2. ASM redundancy and data storage usable capacity

X3-2 Eighth Rack X2-2 and X3-2

Quarter Rack

X2-2 and X3-2
Half Rack

X2-2, X3-2, X2-8, and X3-8
Full Rack

Usable Capacity, High
Performance Disks,
Normal Redundancy

5.15 TB 10.3 TB 25.2 TB 50.4 TB

Usable Capacity, High
Performance Disks,
High Redundancy

3.6 TB 7.2 TB 16.8 TB 33.6 TB

Usable Capacity, High
Capacity Disks,
Normal Redundancy

27 TB 54 TB 126 TB 252 TB

Usable Capacity, High
Capacity Disks,
Normal Redundancy

18 TB 36 TB 84 TB 168 TB

Not ■ the usable capacity numbers in table 5-2 do not account for Fast recovery area storage versus “normal”
database storage, nor do they account for a normally small dBFS disk group. typically, if you elect to configure an Fra
and back up internally to Exadata, Oracle will allocate 60% of your usable capacity for the rECO grid disks. this will
reduce your usable capacity for database files storage to roughly 40% of the numbers displayed in table 5-2 . For example,
for a Full rack configured with high performance disks and normal aSM redundancy, your available capacity for
non-recovery structures (that is, database files) is reduced to approximately 20 tB. please see recipes 5-5, 5-8, and 5-9
to learn more about this topic.

On Exadata, the Oracle ASM external redundancy is not supported. On non-Exadata storage environments,
many clients elect external redundancy and rely on host-based or array-based RAID to provide data protection. With
Exadata the storage servers do not provide RAID protection; data is protected solely with Oracle ASM redundancy.
You can elect to configure ASM disk groups with no redundancy, but this is not recommended for obvious reasons.
Let's discuss disk drive failure probabilities. Referencing studies conducted by Paris/Long in
www.ssrc.ucsc.edu/Papers/paris-storagess06.pdf , we can consider two types of failure probabilities:

• Independent disk drive failures

• Correlated drive failures based on a global defect with a collection of drives. In this case, we

may expected an “accelerated” failure rate, such as that caused by defects from the same batch
of disk drives, escalated temperatures in the drive enclosure or data center, and so forth.