Two or more disks Capacity is the same as the total capacity of the drives in the array No fault tolerance-risk of data loss is proportional to the number.

Two or more disks

Capacity is the same as the total capacity of the drives in the array

No fault tolerance-risk of data loss is proportional to the number of drives in the array

Failure of any drive causes the loss of all data in the array

Very high performance

No overhead for writing fault tolerance data

Used where performance is important and occasional data loss is acceptable

Implemented in some operating systems. Also available in third party software and hardware packages

Can do one write or two simultaneous reads per mirrored pair

Same write transaction rate as single disks

Twice the read transaction rate as single disks

Transfer rate per block is nearly equal to single disks

Capacity is half the total capacity of the drives in the array

No rebuild is necessary if a disk fails. Data is simply copied to the replacement disk

Data TransferData Word

ECC stands for Error Correction Code

Striping is performed at the bit level

High ratio of ECC disks to data disk with smaller word sizes

Data TransferData Word

Extremely high transfer rates possible

Requires hard drives with specialized ECC circuitry

Expensive

Data TransferData Word

Ratio of data disks to ECC disks decreases with increasing transfer rate requirements

Cost is very high - not economical or practical

Data TransferData Word

Relatively simple controller design

Transaction rate is the same as a simple disk at most (with spindle synchronization)

Data TransferData Word

“On the fly” data error correction

Not implemented commercially

All disks in the array must be accessed for every good read and write

High read and write transfer rate (with synchronized spindles)

Synchronized spindles mean only one transaction processed at a time

Transaction rate the same as a single disk at best

Disk failure has low impact on throughput

Good for large sequential data transfers

Not a good solution for random access and small data transfers

Capacity is total capacity of the disks minus the capacity of one disk

All disks in the array must be accessed for every read and write (Parallel access)

Data Transfer

Striping is at the block level, so RAID 4 is somewhat more efficient than RAID 3

High read transaction rate

High aggregate read transfer rate

Data Transfer

Spindles do not need to be synchronized

All drives are not necessarily involved in a read or write (Independent access)

Highest read transaction rate

Medium write transaction rate

Good aggregate transfer rate

Individual block data transfer rate the same as an individual disk

Disk failure has a medium impact throughout

Spindles not synchronized

Data striped at the block level

Data striped on block level

Parity generated just like Raid 5. Second set of parity is calculated and also written across all the drives

Very reliable

Can sustain two simultaneous drive failures

Very high controller overhead to compute parity addresses

Poor write performance

Capacity is N-2 drives

Data Transfer

Implemented as a RAID 0 array whose elements are RAID 1 array

Same level of fault tolerance as RAID 1

Same overhead for fault tolerance as RAID 1

Data Transfer

High I/O rates are achieved by striping RAID 1 arrays

All drives must move in parallel to proper track, lowering sustained performance

Data Transfer

Expensive – capacity is half the total capacity of the disks

Excellent solution for sites that want RAID 1, but need increased performance

Data Transfer

High performance is a result equal distribution of I/O load

Reduces hot spotting in the disks, continued fill of the drive buffers for high sustained transfer rates on large transfers, multiple spindle access for high performance random transactions