Cloud computing platform practices

 

I have not written an article for a while, and soon I graduated from college. After relaxing for a while, I had a pleasant life in my senior year. I 've been relaxed for too long, and I 've got to work again.

It was a little difficult, but I was always pushing myself. Today I couldn't help it, and decided to start writing articles again. Tell everyone that this blog will be updated,

Not ruined.

 

This article briefly records my cloud computing platform practices. The concept of cloud computing has always been mentioned.

But at that time, I stayed at the theoretical stage and did not do anything practical, so I have never written anything about cloud computing, because I have always been

First practice, then summarize, and then write an article. In the first half of my senior year, I took my team to complete the phase II development of a project, and then I got an annoying question in the second half.

Then I started to practice my cloud computing platform. The platform has very few features, because I just played it myself, and the original intention is to implement it for the school.

A cloud storage platform that unifies the data sources of schools, facilitates the future network construction and scientific research of schools, and makes more contributions to schools. But the school is tight

It is difficult to buy a server at a cost, just to consider the status quo. So later, I had to do it myself, not because I had no actual needs.

 

This cloud storage platform is a service-as-a-platform model practice. It provides WebService and APIs for calls by other application systems and implements 20 interfaces because only

It is used for simulation. The data model is simplified and only the core fields are retained. The architecture is based on Hadoop + Hbase + EJB. I learned EJB in my sophomore year, but I have never implemented

First, because there were no projects that must use this technology, and second, considering the technical threshold, it was inconvenient for subsequent maintenance. When implementing the platform

The error cannot be zero Based on the node, because the error is not clear, so I always thought it was a problem with the configuration file, and it had been a problem for some time. Later I found that

The problem of password-free mutual access between linux users is simple, but the error is unclear, which leads to misjudgment.

 

Cloud computing technology is still developing. With the adoption of many companies, it has become mature. Any technology needs to be discovered through practice to solve the problem. Currently

To some extent, cloud storage can replace relational databases, and database transactions and advanced queries can be achieved through third-party components. Loose Data Structure

It is more conducive to agile practices, and data model fields can be changed more flexibly. My cloud storage platform data is to transition some relational data models of the educational administration system.

To a loose structure based on key-value pairs. Some fields with mappings can be simplified to a key-value relationship. Set all links

Data is scattered into loose data, which is indeed Revolutionizing Traditional thinking. However, sometimes reverse thinking is the solution to the bottleneck, such as relational data

It is difficult for a data database to sort data from several terabytes of data within a limited period of time at a low cost and extract top 50 million pieces of data.

 

Hbase provides some mechanisms to cache all data from the distributed file storage layer, especially for data that is written multiple times at a time.

Significant. Although I have never seen twitter's data layer implementation, if I design it, I will write the data that occupies the most storage space at a time.

The Weibo posts and comments I read are partially written into NoSQL. To ensure security, write other private data that is highly secure to MySQL, but I still

Common fields can also be synchronized to NoSQL to improve system performance. This can be seen as the most common practice of cloud storage. Another common type of cloud storage

In practice, the document-based distributed storage stores videos, images, music, and files in key-value pairs. The common features of these media are empty.

Large, one upload, multiple reads. The two can be classified into one category. Even if multiple writes are performed, a small amount of data can be read to a data buffer layer,

To deal with frequent changes, and then regularly and quantitatively persist data. Therefore, the proportion of access does not affect the use of cloud storage. Therefore, I used the Relational Data Model

The practice of completely transforming to a loose type.

 

Using some of the design features of Hbase, we can improve the system performance from the design of a loose table structure. For example, the same part of the family column will be physically

It is stored on a closer node. Therefore, you can set the family column to the table name and the label column to the attribute field in the transition from a relational database to a relaxed database,

In this way, the data of the same table can be physically more compact, so that the system can retrieve the relevant attributes of the same data model more quickly. To define and modify the family column

Hbase is similar to data definition in traditional databases, while the label column can be directly used without definition. This feature enables the table structure field to be dynamic, from

The data persistence layer is dynamic. For properties with key-value ing relationships independent of functions, you can use the family column instead

Store key property name, label column stores key property value, and value column stores value of value attribute. This is a practical strategy for loose model design.

 

The business service layer is separated into two layers. One Layer acts as the data operation layer, responsible for data operations and transaction processing, and the other layer acts as the business service layer, responsible

Business Logic assembly, data verification, and identity verification. The interface design should ensure the rationality, clarity and reusability of the interface. This is what we can think of now.

Some.

Architecture

 

 

Some of my friends are very concerned about my next plan. I would like to give a brief introduction here. I plan to prepare for another exam after graduation.

After I finish, I will find a place to work for a year and continue to study. because the more I learn, the less I feel I will do. So I hope to continue further study and improve. Thank you!

I will keep working hard. We look forward to cooperating and communicating with more technical friends.