当前位置:网站首页>RAID disk array
RAID disk array
2022-07-16 07:40:00 【Free to Superman】
Catalog
RAID Disk display
yes Redundant Array ofIndependent Disks Abbreviation , Chinese abbreviation is Independent redundant disk array
frequently-used RAID Level
RAID0,RAID1,RAID5,RAID6,RAID1+0 etc.
| Disk utilization | Whether the recovery can be verified | Security | Write performance | |
| RAID 0 | N | nothing | nothing ( Only fast ) | On a single disk N times |
| RAID 1 | N/2 | nothing | Allow a device failure | You need to write two storage devices , Prepare for each other |
| RAID 5 | (N-1)/N | Yes | Allow a device failure | Need to calculate and verify |
| RAID 6 | (N-2)/N | nothing | Two equipment failures are allowed | Double write calculation verification is required |
| RAID 1 0 | N/2 | nothing | Allow two base groups to break one each | N/2 The disk is written simultaneously |
RAID 0
RAID 0 Divide data continuously in bits or bytes , Read in parallel / Write on multiple disks , Therefore, it has high data transmission rate , But it has no data redundancy
RAID 0 Just to improve the performance , There is no guarantee for the reliability of the data , And a disk failure will affect all data
RAID0 It can't be used in the situation with high data security requirements
RAID 0 Characteristics
You need at least two disks
Distributed data strip
There is no redundant , Best performance ( Do not store images 、 Check information )
It can't be used in the situation of high data security
RAID 1
Realize data redundancy through disk data mirroring , Generate backup data on a pair of independent disks
When raw data is busy , Data can be read directly from the mirror copy , therefore RAID 1 Can improve read performance
RAID 1 It's the highest cost per unit in a disk array . But it provides high data security and availability . When a disk fails , The system can automatically switch to read and write on the mirror disk , And there's no need to reorganize failed data .
RAID 1 Characteristics
You need at least two disks
Provide data redundancy ( Provide backup )
Good performance
RAID 5
N(N≥3) The disks form an array , A data generation N-1 A strip , There is also a verification data , common N The data is in N Cycle balanced storage on the block disk
N Read and write at the same time , High reading performance , But because of the problem of checking mechanism , Write performance is relatively low
(N-1)/N Disk utilization
High reliability , Allow one disk to break , It doesn't affect all the data
RAID 5 Characteristics
least 3 Block disk
The data is striped
Redundancy with parity
Suitable for reading more and writing less , It's the best compromise between performance and data redundancy
RAID 6
N(N≥4) The disks form an array ,(N-2)/N Disk utilization
And RAID 5 comparison ,RAID 6 A second independent parity information block is added
Two independent parity systems use different algorithms , Even if two disks fail at the same time, data usage will not be affected
be relative to RAID 5 There are bigger ones “ Write loss ”, Therefore, the write performance is poor
RAID 1+0( First make a mirror image , And then make a strip )
N ( even numbers ,N>=4). After two disks are mirrored , Then combine it into a RAID 0
N/2 Disk utilization
N/2 All disks are written simultaneously ,N All disks are read at the same time
High performance , High reliability
RAID 1+0 Characteristics
least 4 Block disk
According to the first RAID 0 Divide into two groups , And then press RAID 1 Mode image
Take care of redundancy ( Provide mirrored storage ) And performance ( The data is striped )
It is commonly used in practical application
experiment
RAID 5
1. First fdisk -l Check the added disks

2. Then divide the four disks into main areas ( The operation of the four disks is the same )

3. After dividing fdisk Check it out. 
4. install mdadm( Already installed can also yum to update )
5. Create and view raid 5
6. View disk details

7. Create a new directory , Mount after formatting

8. Try to make a disk fail , See whether the empty disk can replace the faulty disk 
RAID 0
Divide the main area and install mdadm Reference resources raid5


RAID 1
Divide the main area and install mdadm Reference resources raid5


RAID 1 0
Divide the main area and install mdadm Reference resources raid5



边栏推荐
- Is it reliable to switch to software testing at the age of 30? The mental journey of a person who came here is for you who are confused
- From function test to automatic test, to double the salary, I collated the super complete learning guide [with learning notes]
- 2021-11-13攻防世界做题记录01MISC
- [leetcode] interview question 01.05 Edit once
- Week4
- 【LeetCode】307. Area and retrieval - array modifiable
- 【LeetCode】380. O (1) time insertion, deletion and acquisition of random elements
- Non overlapping interval
- 【剑指Offer】链表专项总结
- 2021-11-13 attack and defense world test record 01misc
猜你喜欢

字典树

2021-11-7bugku做题记录25——POST

help one another in defense work

字节测试总监熬夜10天肝出来的测试岗面试秘籍,给你的大厂梦插上翅膀~

接口测试与接口测试自动化

"Why do you want to resign, Tencent, which broke its head?"
![[sword finger offer] special summary of linked list](/img/5d/7b951a75caf8818cb1191f83b7fa94.png)
[sword finger offer] special summary of linked list

6、 Configuration experiment report of data backup software

2021/12/12 attack and defense world reverse question record

JVM对象创建和内存分配机制
随机推荐
[leetcode] interview question 01.05 Edit once
一位年薪35W的测试被开除,回怼的一番话,令人沉思
Code quality inspection based on sonarqube
2021/12/12 attack and defense world reverse question record
On the meaning of XML file format in the framework
The core difference between fairness and unfairness of reentrantlock
MySQL lock mechanism
IO multiplexing
JVM principle and Practice
Semaphore, countdownlatch use and talking about the source code
Leetcode lecture - 873 Length of the longest Fibonacci subsequence (difficulty: medium)
MySql锁机制
Non overlapping interval
Flask基础入门六-上下文
Démarrer avec les bases de flask VII - cookies et sessions
2年时间,涨薪20k,从外包手工到测试经理的蜕变...
When byte hung up, the interviewer asked me DDD, but I didn't know
01 knapsack filling form implementation
Interviewer: let's talk about how you solve the transaction problem in the distributed scenario?
静态路由的原理和配置