Calculate the usable capacity of RAID, which combines several disks so they act as one, and how well it survives disk failures. Switch between RAID levels to compare the trade-off between capacity efficiency and reliability, and see on a diagram which disks hold data and parity.
Favorites and recently used calculators are stored only in this browser (localStorage).
Result
Usable capacity
–
Capacity efficiency (usable ÷ total)
–
Disk failures tolerated
–
Chance of data loss in a year (estimate)
–
D is data, M is a mirror (a copy of the same data), and P and Q are parity. RAID 5 and 6 rotate the parity disk from stripe to stripe (the diagram shows 4 stripes and up to 12 disks).
Comparison by level
The chance of data loss is a rough, pessimistic estimate assuming each disk fails independently during the year and failed disks are not replaced or rebuilt. In practice failed disks are replaced and rebuilt quickly, so the chance is far smaller, but there is a risk of another failure or an unreadable sector (URE) during the rebuild. RAID is not a substitute for backups.
See it as a graph
Other inputs stay at their current values while the chosen input varies from 1/4 to 4 times its current value. The red dot is the current value.
What is RAID?
RAID (Redundant Arrays of Inexpensive / Independent Disks) combines several disks so they can be used as a single disk. Writing data across several disks (striping) makes it faster, and storing the same data or information that can rebuild it (parity) on other disks (redundancy) keeps data safe when a disk fails. Each level strikes a different balance between usable capacity, speed and reliability.
RAID 0, 1 and 10
RAID 0 only spreads data across all disks: all capacity is usable and it is fast, but if any one disk fails, all data is lost. RAID 1 writes the same data to several disks (mirroring), so only one disk's worth of capacity is usable, but the data survives as long as one disk remains. RAID 10 builds mirrored pairs and stripes across them; it uses half the capacity but combines speed and reliability. It can survive several failures as long as both disks of the same mirror do not fail.
RAID 5, 6 and parity
RAID 5 writes parity (such as the XOR of the data) along with the data and uses one disk's worth of capacity for it. Whichever single disk fails, its contents can be calculated back from the data and parity on the remaining disks. The parity location rotates across disks so that writes do not concentrate on one disk. RAID 6 keeps two kinds of parity and uses two disks' worth of capacity, so it survives two simultaneous failures. With six 4 TB disks, RAID 5 gives 20 TB and RAID 6 gives 16 TB of usable capacity.
Rebuilds and backups
When a disk fails, it is replaced and its contents are rebuilt from the other disks. With large disks a rebuild can take hours or days, and if another disk fails or an unreadable area is found in the meantime, RAID 5 loses data. That is why RAID 6 is increasingly chosen for large disks. RAID also does not help when data is deleted by mistake or encrypted by malware, so it is combined with backups to another location.
AData is divided and written across several disks (striping) to speed up reads and writes, but there is no redundancy
BThe same data is written to two disks (mirroring), so no data is lost if one disk fails
CData and parity for error correction are written distributed across several disks, so data can be restored if one disk fails
Show answer and explanation
Answer: B. The same data is written to two disks (mirroring), so no data is lost if one disk fails
RAID 1 is mirroring, which writes the same data to two disks, so if one disk fails, processing can continue with the data on the other. The capacity available for data is half the total of the two disks. Striping alone with no redundancy is RAID 0, and writing parity distributed across disks is RAID 5. Note that RAID protects against disk failures; data deleted by mistake or encrypted by ransomware is reflected on both disks, so RAID is no substitute for backups.
RAID 5 writes data and parity distributed across all the disks and uses one disk's worth of capacity for parity. The capacity available for data is therefore (4 − 1) × 4 TB = 12 TB, and if one disk fails, the lost data can be restored from the data and parity on the remaining disks. 8 TB is the capacity for RAID 1 or RAID 10, which halve the capacity, and 16 TB is the capacity for RAID 0, which has no redundancy.
RAID 6 writes two kinds of parity distributed across the disks and uses two disks' worth of capacity for parity. The capacity available for data is (6 − 2) × 8 TB = 32 TB, and no data is lost even if up to two disks fail at the same time. As disk capacities grow, recovering (rebuilding) from one disk failure takes longer and the risk of another disk failing in the meantime increases, so RAID 6 is often chosen over RAID 5.
There are C(4, 2) = 6 ways to choose the two failing disks out of four. Data is lost when both disks of the same mirrored pair fail, and since there are two mirrored pairs, that happens in 2 ways. The probability of data loss is therefore 2 ÷ 6 = 1/3. In the remaining 4 ways, one disk remains in each mirrored pair and no data is lost, so RAID 10 can survive two failures depending on the combination, unlike RAID 6, which survives any two.
The four disks together can process 4 × 200 = 800 I/O operations per second. In RAID 5 each write needs four I/O operations, so the number of writes is 800 ÷ 4 = 200 per second. This increase in the I/O operations needed per write is called the write penalty; for mirrored RAID 10 it is 2, giving 400 per second. For write-heavy uses, the RAID level is chosen considering this performance difference as well as capacity efficiency.
FAQ
What is the capacity of RAID 5 with four 4 TB disks?
One disk's worth goes to parity, so (4 − 1) × 4 TB = 12 TB. The capacity efficiency is 75%.
Which is better, RAID 10 or RAID 5?
RAID 10 writes faster and rebuilds quickly but uses half the capacity. RAID 5 uses capacity more efficiently, but calculating parity on every write tends to make writes slower, and the risk during a rebuild is higher. Write-heavy uses such as databases often choose RAID 10.
Do I still need backups if I have RAID?
Yes. RAID protects against disk failures but not against accidental deletion, software bugs, encryption by ransomware or fire. Backups to a separate device or location are needed.
How does this appear on exams?
IT certification exams such as Japan's Fundamental Information Technology Engineer Examination often ask about the features of each RAID level (striping, mirroring, parity), the usable capacity from the number and size of disks, and how many disk failures a configuration can tolerate.