6.155 bytes	Service Hints & Tips	CPES-38LFRJThis document is provided to IBM 
and our Business Partners to help sell and/or service IBM products. It is 
not to be distributed beyond that audience or used for any other purpose.
Server - RAID level descriptions
Applicable to: World-Wide
Service Information:

RAID Level 0

Characteristics:
Data is striped across all drives
Drive spindles need not be synchronised
Number of drives and performance is scalable.

Error Recovery:
No parity information stored on the drives.
Data recovery is limited to the drive electronics
The only way to recover the data is to use system backups

Advantages:
High I/O transaction rate
Multiple simultaneous seeks improve speed
Transparent to system software
Faster in smaller server applications
Load is balanced across all drives

Disadvantages:
No parity or redundancy
No increase in data availability

RAID Level 1

Error Recovery:
Error recovery is through the mirrored drive

Characteristics:
Data duplicated on mirrored drives
Drive spindles are not synchronised
Performance is a function of the number of drives
Load is not balanced across the drives

Advantages:
Maximum data availability
Transparent to system software
No performance penalty after a drive failure

Disadvantages:
100% cost overhead
Requires twice the space and power

RAID Level 2

Error Recovery:
Errors are corrected by recalculating the failed bit from the ECC data 
stored in the array

Characteristics:
Data is spread across all the drives
Drive spindles must be synchrinised
The larger the array, the more efficient the storage capacity of the
array
Data is corrected by the controller on the fly

Advantages:
High data transfer rate
High data security, as data on any single dive is useless
High data availability
Transparent to system software

Disadvantages:
Poor for transaction processing
High cost for small arrays
Special drives/drivers/operating systems required

RAID Level 3

Error Recovery:
Error recovery is through the use of parity information stored on 
dedicated drive
Performed automatically by the disk controller

Characteristics:
Data is spread acroos all the drives in bytes
Drive spindles must be synchronised
Number of drives and performance is scalable
Suited for large file transfers
Suitable for workstations that require fast sequential access to a single
large file

Advantages:
Highest data transfer rate of any RAID level
Increased reliability and data availability

Disadvantages:
Poor performance for small block transfers
Slower than RAID-5 for transaction processing
Unsuitable for a multi-tasking system
May create integration problems [Logical block write size = (number of
drives) * (drive sector size)

RAID Level 4

Error Recovery:
Error recovery is through the use of parity information stored on 
dedicated drive Done automatically

Characteristics:

Data is spread across all the drives in blocks
Drive spindles are not synchronised
Dedicated parity drive
Independant actuators
Number of drives and performance is scalable

Advantages:
Good performance for reads during transaction processing
Multiple actuator seeks improve speed
Increased reliability and data availability

Disadvantages:
Poor performance for write operations
Slower than RAID-5 for transaction processing
Parity disk bottleneck
Performance degradation during data reconstruction

RAID Level 5

Error Recovery:
Error recovery is through the use of parity information distributed across 
multiple drives.
Done automatically by the disk controller

Characteristics:
Data is spread across all the drives in blocks
Drive spindles are not synchronized
Parity is spread across all drives
Independant actuators

Advantages:
Good performance for all transaction processing systems
Multiple actuator seeks improve speed
Increased reliability and data availability
Transparent to system software
Increased reliability with minimal drive cost

Disadvantages:
Performance degradation during data reconstruction

RAID Level 6

Error Recovery:
Error recovery is through the use of mirrored drives

Characteristics:
Data is spread and striped across all the drives in blocks
Data is duplicated across two SCSI buses
Drive spindles are not synchronised
Number of drives and performance is scalable

Advantages:
Highest I/O transaction rateof any RAID level
Multiple actuator seeks improve speed
Maximum data availability
Transparent to system software
No performance penalty after a drive failure
No performance penalty in high write environments

Disadvantages:
An instant 100% cost overhead
Twice the space and power requirements