Disk array basics

xiaoxiao2021-04-03  217

Disk array basics

Disk Array

Disk arrays are combined with two or more hard drives through hardware or management software, and for the server

One or more logical hard drives. The computer can see is the storage space and the remaining free space, and the storage is not available.

How is it stored on a disk. Usually we use SCSI disks to do disk arrays, current Ide devices can also be used

Array technology, but the technical level is still low, so much for low-end clients or low-end servers. From performance angle

It is best not to use multiple SCSI devices in the array. By using HP's disk array card

Various RAID discs are configured, or the partial network operating system can implement the configuration of the array, but because of the soft arrays

Parts requires a large number of CPUs and memory resources, so the software array will reduce the performance of the server.

Why use disk arrays?

When I evaluated a server performance, I often use a concept is the principle of wooden barrel, in South China.

Fang often uses a wooden bucket, a wooden barrel generally uses several wooden board stitching, if this concept is introduced into the service

When the machine can be considered, the components of the computer are the plate of the wooden barrel. If the board of the wooden barrel is high,

The highest water level of this wooden barrel is determined by the lowest wooden board. Therefore, when the performance of the web server, all portions

The lowest performance device determines the highest performance of the server.

So, which equipment is lower?

Now let's analyze the performance parameters of the core components during the running process.

Intel PIII CPU 1000MB / s

SDRAM memory 800MB / s

AGP graphics card 532 ~ 1000MB / s

Hard disk 80MB / s

As can be seen from the list above, the hard disk is the network server bottleneck, and many people have this feeling.

When we use the mouse to click on an application, a small hourglass appears on the screen, the red light of the hard disk keeps flashing, this

Because we need a lot of corresponding programs or data from hard drives, and today's hard disk read and write speed limit us.

The rapid execution of the program. Why is this so? Because today we use the IDE hard drive or ordinary SCSI

Hard disk, only on each of the hard drives inside the computer, the current hard disk theory

You can support the rate of 80MB per second, but during actual operation, it is often not reached, it is better to read and write.

The rate can reach 20MB per second, which is already an ideal indicator, but with CPU and memory IO

It is too low compared to the rate. Is there any solution?

Hardware array

Ordinary SCSI card can only read and write a hard disk at every moment, there is no reading and writing of multiple hard drives at the same time

SCSI control card? Yes, the Netraid array card provided by HP is the performance of excellent performance and powerful.

Column, through HP's array card, you can read and write up to dozens of hard drives at the same time, and ordinary SCSI cards can only provide

512KB cache, while HP's array can configure a cache of 16 ~ 256MB, with traditional SCSI card

Comparison, performance provides several times, and the effect is amazing by actual testing. And the HP array card also provides a hard disk damage time

According to automatic recovery, when the hard disk capacity is not enough, it can be directly charged in a state where the network operating system is run.

A block new hard disk, and the newly inserted hard drive is dynamically added to the original disk array to achieve online capacity expansion. benefit

The popular NetRAID array has a variety of powerful functions, which will not be described in detail.

Through the above introduction, we have a solution that solves the performance of hard drives, in fact, use disk arrays

Technology has another important reason, which is to effectively implement data redundancy through disk arrays. The data can be stored in a mirror or verified manner by disk array technology. Once a block in the disk array has failed. ,

The remaining hard drives can still restore the saved data.

Summary The advantage of disk arrays is:

Maximize IO performance of the disk subsystem

Provide data redundancy for the server

Hardware array composition: array card hard drive

Software array

The software array refers to the common SCSI card that will be connected through the disk management function provided by the network operating system itself.

Multi-block is configured to logic disk, and the software array can provide data redundancy, but the performance of the hard drive is reduced.

Currently Windows NT, NetWare, all operating systems can provide software array features, of which Windows

NT can provide RAID0, RAID1, RAID5. The NetWare operating system can implement the RAID1 function.

Analyze the working principle of the software array to understand the advantages and disadvantages of the software array, when we use the software array

When the hard disk is configured RAID1 or RAID5, the data is stored in a mirror or verification method, and the hard disk has failed.

Storage in the RAID disk still does not lose, guarantees the security of the data. Since the software array uses ordinary SCSI

Card, almost all web servers can use software array configuration, software with network operating system, with

When you set the software array, you can implement it separately, you can implement the same hard drive, so the software array has a low cost, and the configuration is simple.

Unilateral, easy to use.

The disadvantage of the software array is that the software array needs to perform the corresponding disk management program, and the past is for a hard disk.

Operation, and now you need to operate more hard drives, so you need to take up the CPU and internal deposit capital of the web server.

Source, at the same time, in order to ensure the synchronization of the data, the step of adding data verification is required, so the server is on the hard disk system.

The overall performance is more than a single hard drive, and the server also needs additional CPU and memory resources for disk management.

Used, so the overall performance of the server drops approximately 20 ~ 30%. At the same time, the software array does not have a hardware array.

Online expansion, dynamically modify the disk array, automatic data recovery, etc. We said that the software array is used to change

Safety.

Software Array Composition: Server SCSI Card Hard Drive Operating System Array Program

Array card with hardware arrays

Disk array controller

Disk array controller

D2140A NetRAID -1Si disk array card

D5955A Netraid -3Si Disk Array Card

P3410A NetRAID -1M disk array card (Ultra3, 1 channel)

P3411A NetRAID-2M disk array card (Ultra3, 2 channel, 64MB cache)

P3475A NetRAID-2M disk array card (Ultra3, 2 channel, 128MB cache)

D9161A NetRAID-4M disk array card (Ultra3, 4 channel, 64MB cache)

D9351A NetRAID-4M disk array card (Ultra3, 4 channel, 128MB cache)

Physical hard disk

Physical hard drive, Physical drive. HP currently provided has 9GB, 18GB, 36GB, and 72GB of various types of hard drives.

capacity.

Logical hard disk

The logical hard disk is composed of a part or a plurality of physical hard drives of a physical hard disk. By operating system

A physical hard drive can be divided into two logical hard drives, such as C: discs and D: disks. Or make two physical hard drives a bigger

Logic hard drive.

Logical hard drive can be used as part of the physical disc

You can divide a physical hard drive into multiple logical disks and specify the drive letter and capacity. Such a physics

The disc can be divided into multiple drive characters and different capacities.

Logical hard disk can be composed of multiple physical disks

A plurality of physics can be combined together to form a large logical disk.

For operating systems, it cannot distinguish between physical discs and logical disks, which think it is the same. The technology through the logic disk can achieve an increase in disk throughput or disk redundancy.

RAID array technical features

RAID 0 Striping:

Disperse data is prevented from being on each hard disk of the array, each hard disk is stored different data, with maximum data store

Storage capacity; low cost; fast speed, but no data redundancy. RAID0 is a relatively special disk array, usually

Used for not required data to have redundancy, but is required to be used in terms of performance of disk arrays. For example, temporary storage data. save to

The database, for the temporary database that meets high-speed access, etc..

RAID 1 Mirroring / Duplexing:

RAID1 uses two hard drives to mirror, each of the hard disk stored data is the same as the other. When a hard disk appears

When the barrier is broken, the other hard disk continues to work, and the system will continue to operate. With maximum redundancy; fast recovery, but

higher cost. The capacity of the storage capacity is a hard disk. Suitable for operating systems, requiring secure data storage.

RAID 3 (strip distribution dedicated disk check):

RAID3 uses at least three hard disk configurations, stored on one of the hard drives to store dedicated verification data, when a block is

At the time of failure, other hard drives can restore the data of the failed hard disk by verifying the data, due to the use of one

The block hard disk is used as a check, so the hard disk capacity that can actually use is N-1. RAID3 is high availability, cost,

The performance is balanced, and since the verification data is required, the speed is slower, and 3 hard drives or more are required.

RAID 5 (strip technology distributed check):

RAID5 uses at least three hard disk configurations, and the RAID3 is different, RAID5 separates all the verbs of data.

Store on all hard drives, each of the different places of each hard disk stores both data, and also stores check data. When a piece of hard drive

At the time of failure, other hard drives can restore the data of the failed hard disk by verifying the data, because all checksions

Data To use a hard disk capacity as a check, the hard disk capacity that can actually use is N-1. RAID5

Get a balance between high availability, cost, and performance, due to the calculation of the calibration, is slightly slower; but it is slightly more than using RAID 3

Great. Require 3 hard drives or more.

RAID 10 (Striping of Mirrored Arrays):

RAID10 is a special disk array of exclusive access to HP, which is based on RAID1 and RAID0.

Forming. Use four or six, eight hard drives. Every two hard drives are rouid1, and then the resulting low-level array is re-

Combined with RAID0. In this way, RAID10 can be damaged at the same time in most cases, and the two hard drives can continue to work.

Therefore, fault tolerance is greatly improved. Due to the combination of multiple arrays, redundant hard drives are also increased to two, so it is effective

The capacity used is N-2. The capacity is greater than RAID 1, the cost is higher.

RAID 30 (Striping Of Dedicated Parity Arrays):

RAID30 is a special disk array that is exclimated by HP, which is based on RAID3 and RAID0.

Forming. Use 6, 8, 10, 12, 14, 16, 32 blocks. Use 3 or more hard drives to make RAID3,

The resulting low-level array is then recombined as RAID0. In this way, RAID 30 can be in most cases, two hard drives

At the same time, damage can still continue to work, so fault tolerance is greatly improved. Due to multiple array combinations, redundancy is hard

The disc also increases to two, so the capacity for efficient use is N-2. High capacity, high performance, suitable for

In order to retrieve the system, the system is highly used, and the cost is high.

RAID 50 (Striping of Distributed Parity Arrays): RAID50 is a special disk array that is provided by HP, which is based on RAID5 and RAID0.

Forming. Use 6, 8, 10, 12, 14, 16, 32 blocks. Use 3 or more hard drives to make RAID5,

The resulting low-level array is then recombined as RAID0. In this way, RAID50 can be in most cases, two hard drives

At the same time, damage can still continue to work, so fault tolerance is greatly improved. Due to multiple array combinations, redundancy is hard

The disc also increases to two, so the capacity for efficient use is N-2. High capacity, high performance, suitable for

In order to retrieve the system, the system is highly used, and the cost is high. Capacity is greater than RAID RAID 5;

转载请注明原文地址:https://www.9cbs.com/read-131483.html

New Post(0)