Shulou(Shulou.com)06/02 Report--

Computer hardware composition (an important part of the concern of operation and maintenance):

CPU: central processing unit (CPU), which is the computing and control core of a computer. Its main function is to interpret computer instructions and process data in computer software.

The central processing unit mainly includes: arithmetic unit cache memory

(ALU) (cache)

Memory: memory (internal memory) can be used to store data, because it is an electronic component, the data in memory will disappear after power outage, the access speed is fast, and the capacity is relatively small.

Hard disk: the main external memory of a computer, which can be divided into mechanical hard disk and solid state hard disk, which can retain data permanently, but with relatively slow speed and large capacity.

In-depth understanding of hardware knowledge of memory (based on specific cases)

1. Highly concurrent write

The data that needs to be written is first stored in memory, and when the data in memory reaches a certain amount, all the data is written to disk.

Advantages: making use of the speed advantage of memory, it can speed up the writing speed and improve the user experience, which is very suitable for high concurrency.

Disadvantages: memory can only store data temporarily. If power is cut off, all data will be lost.

Solution:

1. The motherboard is equipped with batteries, which can quickly write data to the hard disk after a power outage.

In order to recover

two。 Use (uninterruptible power supply) UPS power supply

3. Select dual-circuit computer room

2 High read

The hot spot data in the hard disk is written into memory, which is convenient for users to read and improve access friendliness.

The external data that needs to be stored is directly stored in the hard disk and then read into memory for access.

Advantages: hot spot data is in memory, users read fast, increase user experience, but also reduce services

The IO pressure of the device.

Tips: writes data to memory, which is called a Buffer.

Read pre-stored data from memory, which is called cache.

3. At present, the IO ratio of most websites is Input: Output = 1: 10, that is, most websites read their content as users.

Mainly, the proportion of writing data is not large, so it should be read at a high level in architecture and optimization.

The program is carried out, and the specific cases are analyzed.

4. Multi-machine cluster memory cache architecture (in the form of software): memcache (pure memory) redis (memory and disk combination)

Hard disk part

1. Classification: mechanical hard disk, solid state hard disk

two。 API type: IDE, SCSI

SAS: the price is more expensive, the rotational speed is relatively higher, and the capacity is smaller at the same price.

SATA: the lowest price, the largest capacity at the same price, and the slow speed.

SSD (solid state drive): the most expensive, the smallest capacity and the fastest at the same price.

3. According to the hard disk selection of the working environment

(1) regular working environment: a good choice for SAS speed, capacity and price.

(2) offline backup or internal working environment: SATA has large capacity and low requirements for hard disk speed.

(3) High concurrency and small amount of data: SDD is fast, but expensive, so it is suitable for extreme production environment.

4.RATD

Disk array (Redundant Arrays of IndependentDisks) means an array of independent disks with honorable capabilities.

Disk array is made up of many cheap disks, which make up a large disk whose capacity is the sum of all disks. The addition effect produced by providing data from individual disks can improve the system performance of the whole disk.

(1) implementation form: hard RAID: realized through raid card, with high price, high efficiency and security

Soft RAID: realized by software simulation, low price, low system efficiency, and

When there is a problem with the software, it will affect the data security.

(2) RAID classification:

RAID 0 striping (Data Stripping) consolidates multiple disks into one disk.

After consolidation, the way data is read from a single disk does not change, but the way data is read from the entire disk array becomes very interesting. First, each disk is split into a number of block at a reasonable size, and each disk must be split the same way. When the data is written, the data is written to three disks in turn, that is, three data blocks [block1]

[block2] [block3], the first disk is written to [1], the second disk to [2], the third to [3], and so on until the data has been written.

When reading data from the disk array, the first block in the disk array was read at the same time, so it took time to read one block and read N block (N is the number of disks in the disk array). In this way, the reading speed of the data is accelerated.

Because RAID 0 only integrates multiple hard drives and does not do redundant processing, then one of the hard drives goes wrong and the whole array will have a problem. Therefore, the reliability of RAID 0 is the same as that of a disk, and the reliability is greatly reduced.

Speed: n times that of a single disk

Capacity: n times the size of a single disk

Reliability: 1Compact N of a single disk

Applicable scenarios: only for those scenarios that do not require high data security

RAID 1 mirroring mirrors data from one disk to another as redundancy

In RAID 1, when data is written to one disk, it is also written to another disk as a mirror. When there is a problem with the disk being used, the system will abandon it and go to its mirrored disk to read the data, which can be replaced by the broken disk. In this way, the security of the data is greatly increased, but the cost is also increased, and the disk utilization is 50%, which is reduced by half.

After replacing the disk, the system will synchronize the original data, which will affect the IO efficiency of the system during the synchronization period.

Speed: constant

Capacity: 50% of all disks

Reliability: greatly increased

Usage scenario: the data is important and the amount of data is small.

Independent disk structure of RAID 5 distributed parity

RAID 5 is similar in structure to RAID 0 in that it stripes the disk and has a high read speed. When writing data, parity information is written on each disk, resulting in write losses. The data is checked for errors when reading information from the disk array. When a disk goes wrong, after replacing a new disk, its data can be recovered through calculation through the data and parity information of other disks.

Speed: the sum of N disks

Capacity: slightly less than N disks

Reliability: when an error occurs on a disk, it can be repaired: there is a parity code, which can be corrected

RAID 10 (1-0) mirrors first, then stripes

RAID 10 is the combination product of RAID 1 and RAID 0. Each disk is mirrored (RAID 1), and then each formed RAID 1 is striped (RAID 0). Overall, the entire disk array is in the form of RAID 0, and each disk is mirrored. Then it not only has the high reading speed of RAID 0, but also has the high data security of RAID 1. The shortcomings of RAID 10 are also obvious, such as low disk utilization, resulting in higher prices.

Speed: the sum of N hard drives

Capacity: 50%

Reliability: with the support of RAID 0, it can be highly

5. Main brands of PC servers in enterprises

DELL (most enterprises), HP,IBM (it is estimated that no one will use it again when going to IOE)

DELL PC server model

2010-2013 1U R410 R610

2u R710

2014-2015 rig 1u R420 R430 R620 R630

2u R720

For this part of the content I will pay attention to later in the process, "large-scale website technical architecture analysis", "Taobao these 10 years" these two books will be my reference books.

Summary of learning thought

In the university, I majored in software engineering, as you know, which is a science major, but I think it is more appropriate to make it an engineering discipline. Learning the C language, you remember the concepts of variables, pointers, functions and arrays very well. If you can't understand them thoroughly, it's still useless. Just understand the written knowledge, you still can't do anything when you start programming.

Therefore, practice is the most meaningful. When you encounter problems in practice, you have to think about where the problems appear, combined with book knowledge. After that, share and discuss. Only in this way can you deepen the breadth and width of your only, which is in your best interest.

What we have to overcome in the process of learning is laziness. Laziness is not a disease, but your slackness to your ideals and your irresponsibility for the future. Now that you have chosen, you have to stick to it. There is no other way.

My interest is in the Linux part, and I have a strong interest in the background system architecture of the server, so I have reason to believe that I can do it. It is very important to have confidence in yourself.

At this point, I would also like to thank my operating system teacher, Dr. Cai Qingsong. I learned all this writing from him, and Linux learned about it through it. As he said: "Today's students are shit, do not know anything" this sentence is actually the best incentive for us, since not, learn!

It seems that this passage is not like a summary of learning thoughts, it seems that it is not very formal. At that time, I felt that to express my thoughts, I had to say what I thought. If there are really rules and regulations, such as hard work, persistence, confidence and so on, it will be meaningless.

The Development of UNIX and LINUX

Unix is written in a mixture of 90% C language and 10% assembly language, so the code still needs to be changed when porting the hardware platform, so various companies have launched Unix systems for their own models.

When it came to the seventh edition of AT&TSystem V, Unix for X86 was finally released, so it was finally possible to install Unix on personal computers, but there was a rule that "the source code cannot be disclosed to students", so Professor Tanebaum did not look at the Unix source code and wrote a Unix-compatible Minix operating system for X86.

4. Because Minix is only used for teaching, its function is not strong, so Torvalds uses GNU's bash as the development environment and gcc as the compilation tool to write the Linux kernel-v0.02. But at the beginning, Linux is not compatible with Unix, that is, applications running on Unix cannot run on Linux, that is, the interface between the application and the kernel is inconsistent, because Unix follows the POSIX specification, so Torvalds modifies Linux and follows POSIX (PortableOperating System Interface). He standardizes the interface between the application and the kernel)

At first, Linux is only suitable for 386. later, with the help of netizens all over the world, it is finally compatible with a variety of hardware.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.