Introduction to RAID

http://arch.pconline.com.cn/diy/front/evalue/storage/0407/409139_1.html 理论上的双倍性能：RAID-0
　　对于那些早已熟悉RAID以及它怎么运作的人，可以跳过这里直接进入基准测试。下面是对桌面系统上两种最普通的RAID形式做一个简短的介绍：RAID-0和RAID-1。
　　我们在这篇文章中将要谈到，RAID-0是唯一的RAID性能增强形式。它的原理很简单。要被写入驱动器的数据被分割成“段”，通常大小在16-256KB，每一段被写入阵列中的一个不同的驱动器。例如，我们正在操作一个双硬盘RAID-0阵列，段大小为128KB，而我们要写入256KB的数据；硬盘0将接受前128KB数据写入，而硬盘1获得剩下的128KB。

//  

向单个硬盘写入//向双硬盘RAID-0阵列写入

　　从这里你可以看到，RAID-0的写入性能应该差不多是单个硬盘的双倍，因为在相同时间下可以写入两倍那么多的数据。由于RAID控制器必须在发送数据给硬盘之前操作分割数据成段，更高的写入性能是以一些控制器开销为代价的 - 但由于现在的微处理器已经很快了，这个开销通常可以忽略不记。

　　读取以完全相同的方法反向进行。如果我们要读回同一个256KB的数据，就从硬盘0取出一段，而硬盘1取出另一段。现在读取也在一半的时间内完成了，理论上性能翻倍。
　　我们小心地使用了“理论上”这个词，因为如果不是在像我们刚刚描述的理想情况下操作，RAID-0的性能优势会很快消失。如果使用的段大小太大，RAID-0可能会丧失性能优势，而太小的段大小会导致过度的开销，降低阵列的性能改进。
　　在过去我们已经看到，对大多数桌面应用来说，桌面RAID控制器会提供的最大段大小通常是基于性能的最佳选择。Intel的ICH5/6传输在128KB的段大小下，经过比较是最适合的了。其它的段大小选项并不能为我们的测试套件提供任何更好的性能。
　　RAID-0的主要缺陷，除了开销以外，还有可靠性。RAID-0阵列的容量是它的所有成员之和；那么，RAID-0阵列中的两个100GB硬盘将构成一个总容量为200GB的阵列。不幸的是，如果你丢失了阵列中的任何一个硬盘，你所有的数据都会丢失并且不可恢复。由于两个硬盘在协力地工作并对于保存数据都很重要，所以在移动数据到一个双硬盘RAID-0阵列时，你将承担双倍的风险。
把冗余放进RAID：RAID-1
　　我们将要谈论RAID的另一个类型是RAID-1，通常理解为镜像。这里我们不准备对RAID-1进行基准测试，因为在很大程度上，不会有性能的提升或减少。就像标题所指的那样，RAID-1是用于冗余的。

向双硬盘RAID-1阵列写入

　　与RAID-0不同，在数据发送到硬盘以前没有对它的预处理。使用RAID-1，写入到硬盘0的所有东西的副本被写入它的镜像硬盘。RAID-1的好处是如果一个硬盘坏了，你有一个能够提取的完美工作备份，直到你更换了坏掉的硬盘。通过使用双硬盘RAID-1阵列，你有效地把硬盘可靠性翻倍了。你将注意到这跟RAID-0是正好相反的，但RAID-1的缺陷是你使用了两倍的硬盘，而没有获得任何额外的容量或性能，只有可靠性。

何为matrix raid？就是将两个硬盘组成的raid分割成两个，无论是raid0+1还是raid0+0都是matrix raid

至于你的问题是，做raid0+0的话和raid0完全没有区别，区别只是测试好看点而已，实际没有任何提升

AID Overview

This chapter describes RAID (Redundant Array of Independent Disks), RAID functions and benefits, RAID components, RAID levels, and configuration strategies.

This chapter contains the following sections:

•

Information
About RAID

•

RAID
Levels

•

Generic
Drive Replacement Procedure

•

Platform-Specific
RAID and Drive Procedures

Information About RAID

RAID is an array, or group, of multiple independent physical drives that provide high performance and fault tolerance. A RAID drive group improves input/output (I/O) performance and reliability.
The RAID drive group appears to the host computer as a single storage unit or as multiple virtual units. I/O is expedited because several drives can be accessed simultaneously.

RAID drive groups improve data storage reliability and fault tolerance compared to single-drive storage systems. Data loss resulting from a drive failure can be prevented by reconstructing missing
data from the remaining drives. RAID improves I/O performance and increases storage subsystem reliability.

RAID levels describe a system for ensuring the availability and redundancy of data stored on large disk subsystems. SeeRAID
Levels for detailed information about RAID levels. The RAID drive-group components and RAID levels are described in the following sections.

Drive Group

A drive group is a group of physical drives. These drives are managed in partitions known as virtual drives.

Virtual Drive

A virtual drive is a partition in a drive group that is made up of contiguous data segments on the drives. A virtual drive can consist of an entire drive group, more than one entire drive group,
a part of a drive group, parts of more than one drive group, or a combination of any two of these conditions.

Disk Striping

Disk striping (used in RAID level 0) allows you to write data across multiple drives instead of only one drive. Disk striping involves partitioning each drive storage space into stripes that can
vary in size from 8 KB to 1024 KB. These stripes are interleaved in a repeated sequential manner. The combined storage space is composed of stripes from each drive. We recommend that you keep stripe sizes the same across RAID drive groups.

For example, in a four-disk system using only disk striping, segment 1 is written to disk 1, segment 2 is written to disk 2, and so on (see Figure 1-1).
Disk striping enhances performance because multiple drives are accessed simultaneously, but disk striping does not provide data redundancy

Figure 1-1 Example of Disk Striping (RAID 0)



Stripe width is the number of drives involved in a drive group where striping is implemented. For example, a four-disk drive group with disk striping has a stripe width of four.

The stripe size is the length of the interleaved data segments that the RAID controller writes across multiple drives, not including parity drives. For example, consider a stripe that contains
64 KB of disk space and has 16 KB of data residing on each disk in the stripe. In this case, the stripe size is 64 KB and the strip size is 16 KB.

The strip size is the portion of a stripe that resides on a single drive.

Disk Mirroring (RAID 1 and RAID 10)

With disk mirroring (used in RAID 1 and RAID 10), data written to one drive is simultaneously written to another drive. The primary advantage of disk mirroring is that it provides 100 percent data
redundancy. Because the contents of the disk are completely written to a second disk, data is not lost if one disk fails. In addition, both drives contain the same data at all times, so either disk can act as the operational disk. If one disk fails, the contents
of the other disk can be used to run the system and reconstruct the failed disk.

Disk mirroring provides 100 percent redundancy but is expensive because each drive in the system must be duplicated (see Figure 1-2).

Figure 1-2 Example of Disk Mirroring (RAID 1)

Parity

Parity generates a set of redundancy data from two or more parent data sets. The redundancy data can be used to reconstruct one of the parent data sets in the event of a drive failure. Parity data
does not fully duplicate the parent data sets, but parity generation can slow the write process. In RAID, this method is applied to entire drives or stripes across all of the drives in a drive group. There are two types of parity:

•

Dedicated parity—The parity data on two or more drives is stored on an additional disk.

•

Distributed parity—The parity data is distributed across more than one drive in the system.

RAID 5 combines distributed parity with disk striping (see Figure 1-3).
If a single drive fails, it can be rebuilt from the parity and the data on the remaining drives. RAID 5 uses parity to provide redundancy for one drive failure without duplicating the contents of entire drives. RAID 6 uses distributed parity and disk striping
also but adds a second set of parity data so that it can survive up to two drive failures.



Note

Parity is distributed across all drives in the drive group.

Figure 1-3 Example of Distributed Parity (RAID 5)

Disk Spanning

Disk spanning allows multiple drives to function like one big drive. Spanning overcomes lack of disk space and simplifies storage management by combining existing resources or adding relatively
inexpensive resources. For example, four 20-GB drives can be combined to appear to the operating system as a single 80-GB drive.

Spanning alone does not provide reliability or performance enhancements. Spanned virtual drives must have the same stripe size and must be contiguous. In Figure 1-4,
RAID 1 drive groups are turned into a RAID 10 drive group.



Note

Make sure that the spans are in different backplanes, so that if one span fails, you do not lose the whole drive group.

Figure 1-4 Example of Disk Spanning



Spanning two contiguous RAID 0 virtual drives does not produce a new RAID level or add fault tolerance. It does increase the capacity of the virtual drive and improves performance by doubling the
number of physical disks.

Table 1-1 describes
how to configure RAID 00, RAID 10, RAID 50, and RAID 60 by spanning. The virtual drives must have the same stripe size and the maximum number of spans is eight. The full drive capacity is used when you span virtual drives; you cannot specify a smaller drive
capacity.

Table 1-1 Spanning for RAID 00, RAID 10, RAID 50, and RAID 60

RAID Level	Description
00	Configure RAID 00 by spanning two contiguous RAID 0 virtual drives, up to the maximum number of supported devices for the controller.
10	Configure RAID 10 by spanning two contiguous RAID 1 virtual drives, up to the maximum number of supported devices for the controller. RAID 10 supports a maximum of eight spans. You must use an even number of drives in each RAID virtual drive in the span. The RAID 1 virtual drives must have the same stripe size.
50	Configure RAID 50 by spanning two contiguous RAID 5 virtual drives. The RAID 5 virtual drives must have the same stripe size.
60	Configure RAID 60 by spanning two contiguous RAID 6 virtual drives. The RAID 6 virtual drives must have the same stripe size.

Hot Spares

A hot spare is an extra, unused drive that is part of the disk subsystem. It is usually in standby mode, ready for service if a drive fails. If a drive used in a RAID virtual drive fails, a hot
spare automatically takes its place and the data on the failed drive is rebuilt on the hot spare. Hot spares can be used for RAID levels 1, 5, 6, 10, 50, and 60.

Hot spares permit you to replace failed drives without system shutdown or user intervention. MegaRAID SAS RAID controllers can implement automatic and transparent rebuilds of failed drives using
hot spare drives, providing a high degree of fault tolerance and zero downtime.



Note

When running RAID 0 and RAID 5 virtual drives on the same set of drives (a sliced configuration), a rebuild to a hot
spare cannot occur after a drive failure until the RAID 0 virtual drive is deleted.

The LSI RAID management software allows you to specify drives as hot spares. When a hot spare is needed, the RAID controller assigns the hot spare that has a capacity closest to and at least as
great as that of the failed drive to take the place of the failed drive. The failed drive is removed from the virtual drive and marked ready awaiting removal once the rebuild to a hot spare begins. You can make hot spares of the drives that are not in a RAID
virtual drive.

You can use the RAID management software to designate the hot spare to have enclosure affinity, which means that if drive failures are present on a split backplane configuration, the hot spare
is used first on the backplane side that it resides in.

If the hot spare is designated as having enclosure affinity, it attempts to rebuild any failed drives on the backplane that it resides in before rebuilding any other drives on other backplanes.



Note

If a rebuild to a hot spare fails for any reason, the hot spare drive is marked as failed. If the source drive fails,
both the source drive and the hot spare drive is marked as failed.

There are two types of hot spares:

•

Global hot spare

•

Dedicated hot spare

Global Hot Spare

A global hot spare drive can be used to replace any failed drive in a redundant drive group as long as its capacity is equal to or larger than the capacity of the failed drive. A global hot spare
defined on any channel should be available to replace a failed drive on both channels.

Dedicated Hot Spare

A dedicated hot spare can be used to replace a failed drive only in a chosen drive group. One or more drives can be designated as a member of a spare drive pool. The most suitable drive from the
pool is chosen for failover. A dedicated hot spare is used before one from the global hot spare pool.

Hot spare drives can be located on any RAID channel. Standby hot spares (not being used in RAID drive group) are polled every 60 seconds at a minimum, and their status is made available in the
drive group management software. RAID controllers offer the ability to rebuild with a disk that is in a system, but not initially set to be a hot spare.

When using hot spares, observe the following guidelines:

•

Hot spares are used only in drive groups with redundancy, which includes RAID levels 1, 5, 6, 10, 50, and 60.

•

A hot spare connected to a specific RAID controller can be used to rebuild a drive that is connected to the same controller only.

•

You must assign the hot spare to one or more drives through the controller BIOS or use drive group management software to place it in the
hot spare pool.

•

A hot spare must have free space equal to or greater than the drive it replaces. For example, to replace an 18-GB drive, the hot spare must
be 18 GB or larger.

Disk Rebuilds

When a drive in a RAID drive group fails, you can rebuild the drive by recreating the data that was stored on the drive before it failed. The RAID controller recreates the data using the data stored
on the other drives in the drive group. Rebuilding can be done only in drive groups with data redundancy, which includes RAID 1, 5, 6, 10, 50, and 60 drive groups.

The RAID controller uses hot spares to rebuild failed drives automatically and transparently, at user-defined rebuild rates. If a hot spare is available, the rebuild can start automatically when
a drive fails. If a hot spare is not available, the failed drive must be replaced with a new drive so that the data on the failed drive can be rebuilt.

The failed drive is removed from the virtual drive and marked ready awaiting removal when the rebuild to a hot spare begins. If the system goes down during a rebuild, the RAID controller automatically
restarts the rebuild after the system reboots.

RAID 0

RAID 0 provides disk striping across all drives in the RAID drive group. RAID 0 does not provide any data redundancy but does offer the best performance of any RAID level. RAID 0 breaks up data
into smaller segments and stripes the data segments across each drive in the drive group. The size of each data segment is determined by the stripe size. RAID 0 offers high bandwidth.



Note

RAID level 0 is not fault tolerant. If a drive in a RAID 0 drive group fails, the whole virtual drive (all drives associated
with the virtual drive) will fail.

By breaking up a large file into smaller segments, the RAID controller can use both SAS drives and SATA drives to read or write the file faster. RAID 0 involves no parity calculations to complicate
the write operation, which makes RAID 0 ideal for applications that require high bandwidth, but do not require fault tolerance. Table 1-4 provides
an overview of RAID 0.Figure 1-5 shows an example of a RAID 0 drive group
advantage.

Table 1-4 RAID 0 Overview

Feature	Description
Uses	Provides high data throughput, especially for large files. Any environment that does not require fault tolerance.
Benefits	Provides increased data throughput for large files. No capacity loss penalty for parity.
Limitations	Does not provide fault tolerance or high bandwidth. All data is lost if any drive fails.
Drives	1 to 32.

Figure 1-5 RAID 0 Drive Group Example

RAID 1

In RAID 1, the RAID controller duplicates all data from one drive to a second drive in the drive group. RAID 1 supports an even number of drives from 2 to 32 in a single span. RAID 1 provides complete
data redundancy but at the cost of doubling the required data storage capacity. Table 1-5 provides
an overview of RAID 1. Figure 1-6 shows an example of a RAID 1 drive
group.

Table 1-5 RAID 1 Overview

Feature	Description
Uses	Use RAID 1 for small databases or any other environment that requires fault tolerance, but small capacity.
Benefits	Provides complete data redundancy. RAID 1 is ideal for any application that requires fault tolerance and minimal capacity.
Limitations	Requires twice as many drives. Performance is impaired during drive rebuilds.
Drives	2 to 32 (must be an even number of drives).

Figure 1-6 RAID 1 Drive Group Example

RAID 5

RAID 5 includes disk striping at the block level and parity. Parity is the property of the data of being odd or even, and parity checking is used to detect errors in the data. In RAID 5, the parity
information is written to all drives. RAID 5 is best suited for networks that perform a lot of small input/output (I/O) transactions simultaneously. RAID 5 provides data redundancy, high read rates, and good performance in most environments. It also provides
redundancy with the lowest loss of capacity.

In addition, RAID 5 is good for any application that has high read request rates but has low write request rates.

RAID 5 addresses the congestion issue for random I/O operations. Because each drive contains both data and parity, numerous writes can take place concurrently.

Table 1-6 provides
an overview of RAID 5. Figure 1-7 shows an example of a RAID 5 drive
group.

Table 1-6 RAID 5 Overview

Features	Description
Uses	Provides high data throughput, especially for large files. Use RAID 5 for transaction processing applications because each drive can read and write independently. If a drive fails, the RAID controller uses the parity drive to recreate all missing information. Use also for office automation and online customer service that requires fault tolerance. Use for any application that has high read request rates but low write request rates.
Benefits	Provides data redundancy, high read rates, and good performance in most environments. RAID 5 provides redundancy with the lowest loss of capacity.
Limitations	Not well-suited to tasks that require a large number of writes. RAID 5 has problems if no cache is used (clustering). The drive's performance is reduced if a drive is being rebuilt. Environments with few processes do not perform as well because the RAID overhead is not offset by the performance gains in handling simultaneous processes.
Drives	3 to 32.

Note

Parity is distributed across all drives in the drive group.

RAID 00

A RAID 00 drive group is a spanned drive group that creates a striped set from a series of RAID 0 drive groups. RAID 00 does not provide any data redundancy, but along with RAID 0, RAID 00 offers
the best performance of any RAID level. RAID 00 breaks up data into smaller segments and stripes the data segments across each drive in the drive groups. The size of each data segment is determined by the stripe size. RAID 00 offers high bandwidth.



Note

RAID level 00 is not fault tolerant. If a drive in a RAID 0 drive group fails, the whole virtual drive (all drives associated
with the virtual drive) fails.

By breaking up a large file into smaller segments, the RAID controller can use both SAS drives and SATA drives to read or write the file faster. RAID 00 involves no parity calculations to complicate
the write operation, which makes RAID 00 ideal for applications that require high bandwidth but do not require fault tolerance. Table 1-8 provides
an overview of RAID 00. Figure 1-9 shows an example of a RAID 00 drive
group.

Table 1-8 RAID 00 Overview

Features	Description
Uses	Provides high data throughput, especially for large files. Use RAID 00 in any environment that does not require fault tolerance.
Benefits	Provides increased data throughput for large files. RAID 00 has no capacity loss penalty for parity.
Limitations	Does not provide fault tolerance or high bandwidth. All data is lost if any drive fails.
Drives	2 to 256.

Figure 1-9 RAID 00 Drive Group Example Using Two Drives

RAID 10

RAID 10 is a combination of RAID 0 and RAID 1 and consists of stripes across mirrored drives. RAID 10 breaks up data into smaller blocks and mirrors the blocks of data to each RAID 1 drive group.
The first RAID 1 drive in each drive group then duplicates its data to the second drive. The size of each block is determined by the stripe size parameter, which is set during the creation of the RAID set. The RAID 1 virtual drives must have the same stripe
size.

Spanning is used because one virtual drive is defined across more than one drive group. Virtual drives defined across multiple RAID 1 level drive groups are referred to as RAID level 10, (1+0).
Data is striped across drive groups to increase performance by enabling access to multiple drive groups simultaneously.

Each spanned RAID 10 virtual drive can tolerate multiple drive failures, as long as each failure is in a separate drive group. If there are drive failures, less than the total drive capacity is
available.

Configure RAID 10 by spanning two contiguous RAID 1 virtual drives, up to the maximum number of supported devices for the controller. RAID 10 supports a maximum of eight spans with a maximum of
32 drives per span. You must use an even number of drives in each RAID 10 virtual drive in the span.



Note

Other factors, such as the type of controller, can restrict the number of drives supported by RAID 10 virtual drives.

Table 1-9 provides
an overview of RAID 10. In Figure 1-10, virtual drive 0 is created by
distributing data across four RAID 1 drive groups (drive groups 0 through 3).

Table 1-9 RAID 10 Overview

Benefits	Description
Uses	Appropriate when used with data storage that needs 100 percent redundancy of mirrored drive groups and that also needs the enhanced I/O performance of RAID 0 (striped drive groups.) RAID 10 works well for medium-sized databases or any environment that requires a higher degree of fault tolerance and moderate to medium capacity.
Benefits	Provides both high data transfer rates and complete data redundancy.
Limitations	Requires twice as many drives as all other RAID levels except RAID 1.
Drives	4—The maximum number of drives supported by the controller (using an even number of drives in each RAID 10 virtual drive in the span).

Figure 1-10 RAID 10 Virtual Drive Example

RAID 50

RAID 50 provides the features of both RAID 0 and RAID 5. RAID 50 includes both parity and disk striping across multiple drive groups. RAID 50 is best implemented on two RAID 5 drive groups with
data striped across both drive groups.

RAID 50 breaks up data into smaller blocks and stripes the blocks of data to each RAID 5 disk set. RAID 5 breaks up data into smaller blocks, calculates parity, and writes the blocks of data and
parity to each drive in the drive group. The size of each block is determined by the stripe size parameter, which is set during the creation of the RAIDset.

RAID level 50 can support up to eight spans and tolerate up to eight drive failures though less than total drive capacity is available. Though multiple drive failures can be tolerated, only one
drive failure can be tolerated in each RAID 5 level drive group.

Table 1-10 provides
an overview of RAID 50. In Figure 1-11, virtual drive 0 is created by
distributing data across two RAID 5 drive groups.

Table 1-10 RAID 50 Overview

Features	Description
Uses	Appropriate when used with data that requires high reliability, high request rates, high data transfer, and medium to large capacity.
Benefits	Provides high data throughput, data redundancy, and very good performance.
Limitations	Requires 2 to 8 times as many parity drives as RAID 5.
Drives	Eight spans of RAID 5 drive groups containing 3 to 32 drives each (limited by the maximum number of devices supported by the controller).

Figure 1-11 RAID 50 Virtual Drive Example

Add to span 什么意思

我其实已经很久没有配置过服务器的阵列。来这公司的时候，发现配置阵列的界面已经完全改变，很多东西都不是那么熟悉。

其实那些配置的英文，如果我认真看看，基本都能搞明白。不过有些地方是怎么看都不明白。
http://bbs.hdchina.org/viewthread.php?tid=42922 RAID技术中SPAN选项RAID级别

RAID 0

RAID 0对应RAID级别中最高的存储性能，其原理是把连续的数据分散到多个磁盘上存取。RAID 0的缺点是不提供数据冗余，数据可靠性差。
RAID 1

RAID 1又称为Mirror或 Mirroring，它的宗旨是最大限度的保证用户数据的可用性。 RAID 1的操作方式是把用户写入硬盘的数据百分之百地自动复制到另外一个硬盘上。
RAID 10

正如其名字一样，RAID 10是RAID 0和RAID 1的组合形式。

“SPAN”的多重意义

SPAN作为一种RAID模式

i. 有一种RAID BIOS中，会将RAID SPAN与RAID0、RAID 1、RAID 10等并列。在这种情况下，SPAN被理解为一种RAID模式：JBOD（Just Bundle OfDisks），译成中文可以是“简单磁盘捆绑”，又称作SPAN。

ii. 与RAID 0一样，Span也不提供数据安全保障；Span的存储容量等于组成Span的所有磁盘的容量的总和。

iii. 与RAID0不同的是，Span上的数据简单的从第一个磁盘开始存储， 当第一个磁盘的存储空间用完后， 再依次从后面的磁盘开始存储数据。因此，Span的存取性能完全等同于对单一磁盘的存取操作。

iv. SPAN不是标准的 RAID级别，只是在近几年被一些厂家提出，并被广泛采用。
SPAN作为一项RAID选项

i. 在一些RAID BIOS设置中，有时也会看到SPAN选项，此时SPAN意为“阵列跨接”。将SPAN选项设置为CANSPAN，使得逻辑驱动器可以跨接多个物理阵列。

ii. 阵列跨接可以用来实现RAID 10/RAID 50。比如，在DELL的一些机型中就是通过跨接两个RAID 1来实现RAID 10。
其它情况

另外，在少数一些资料中，SPAN被归为RAID 0的一种类别。