7.3. RAID Implementation

Two different levels of RAID technology exist: software-based RAID and hardware-based RAID.

In software-based RAID, the array management functions are implemented by software executing in a host environment, such as the operating system.

With hardware-based RAID, the array management functions are implemented by hardware and firmware within a RAID controller. The RAID controller orchestrates read and write activities in the same way a controller for a single disk drive does, and treats the array as if it were a single or virtual drive. This level is transparent to the host software and is self-contained within the storage solution. It does not interfere with the primary function of the host, which translates into higher overall performance and potential for advanced RAID capabilities.

Figure 7-3 contrasts software RAID and hardware RAID.

Figure 7-3. Software-based RAID vs. hardware-based RAID.

Note

The ability to use online spares is available with hardware-based RAID. Online spares are extra drives that are automatically brought online when a drive fails in a RAID set.

The advantages of software-based RAID are as follows:

• The ability to span multiple storage subsystems or even hosts

• Lower cost

The disadvantages of software-based RAID are as follows:

• Lower performance than hardware RAID

• No boot capability from RAID 5 protected arrays (Microsoft Windows)

• No support for RAID 1+0 or RAID ADG

• No support for RAID 5 support under Novell NetWare

• No support for PCI Hot Plug for RAID 5 protected arrays (Windows)

• No support for online spare drives

• Increased processor load

• Configuration utilities not user-friendly

• Operating system interface required when starting a rebuild

In a software RAID solution, each sector requires an extra transfer of three sectors across the I/O bus. This requires 256 transfers on the I/O bus. The memory bus and the processor bus also have to execute extra cycles for parity generation.

After a drive failure, a hardware protected RAID 5 logical drive is several times faster than a software-protected drive, due to the overhead of reconstructing data of the failed drive from the remaining drives.

In a hardware RAID solution, after the data has reached the array controller there is no extra activity on the I/O bus, the memory bus, or the frontside bus. All RAID-related activities are local to the array SCSI bus. For a single sector that is written to the disk, 64 transfers are executed on a 64-bit PCI bus.

The advantages of HP hardware-based RAID are as follows:

• Safe write caching through the array accelerator

• Superior manageability, such as multiple volumes with different RAID levels

• High flexibility such as separation of I/O into different disk volumes

• Hardware RAID, automatic recovery, PCI Hot Plug capability, and online spares

• Automatic data distribution and I/O balancing across multiple disk drives

• Superior performance and I/O optimization

7.3.1 Mean Time Between Failure

Mean time between failure (MTBF) is the expected time after the initial burn-in phase that it takes a hardware component to fail because of normal wear and tear. For disk drives, MTBF is calculated from the theoretical steady state failure rate from a large population of drives tested in volume manufacturing.

The actual individual drive MTBF depends on the drive usage and environmental conditions. Stressing a drive beyond normal usage can significantly reduce the predicted MTBF of the drive.

The MTBF of an array is equal to the MTBF of an individual drive divided by the number of drives in the array. The number lowers because there are more physical spindles that are subject to failure. This is a good reason to use RAID configurations that support fault tolerance.

Example

If the MTBF of a single drive is 200,000 hours, the MTBF of an array with 5 similar drives is calculated as 200,000 divided by 5, for a total array MTBF of 40,000.

It is important to note that as the number of drives increases in a RAID configuration, the efficiency of the array increases but does not change the MTBF of each drive predicted by the drive manufacturer.

7.3.1.1 EXTENDING MTBF

To offset the increased probability of drive failure resulting from increasing the number of drives by using drive arrays, an online spare can be added to any of the fault-tolerant RAID levels. An online spare further decreases the probability of logical drive failure to about a thousandth of the previous level.

In addition, HP ProLiant servers provide full-spectrum fault management. ProLiant servers employ a unique set of high-availability technologies for handling faults and providing fault prevention, fault tolerance, and rapid recovery.

7.3.2 Striping Factor

Striping unites multiple physical drives into a single logical drive. The logical drive is arranged so that blocks of data are written alternately across all physical drives in the logical array. The number of sectors per block is referred to as the striping factor.

Depending on the array controller in use, the striping factor can be modified, usually with the manufacturer's system configuration utility. Many of the Smart Array controllers can be modified with the Array Configuration Utility (ACU).

7.3.3 RAID Levels

Industry-standard RAID levels are listed in the following table.

Level	Description
RAID 0	Data striping without parity
RAID 1	Disk mirroring
RAID 1	Disk duplexing
RAID 2	Complex error correction
RAID 3	Parallel-transfer, parity drive
RAID 4	Concurrent access, dedicated parity drive (data guarding)
RAID 5	Concurrent access, distributed parity (distributed data guarding)
RAID 1+0	Disk mirroring and data striping without parity
RAID ADG	Distributed data guarding with two sets of parity

Note

RAID levels 2 and 3 are no longer used in the industry.