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

Driven by new technologies such as cloud computing and big data, more and more enterprises need to query, analyze, process and update structured data. At the same time, with the continuous increase of innovative business, the complexity and huge volume of business will inevitably produce complex and large-scale structured data, which will inevitably force enterprises to point to large-scale, high-reliability, high-scale and high-performance database.

For any enterprise, database is the core resource of its application. How to choose the core database suitable for its own enterprise IT architecture and business is a problem that every CIO/CTO must face.

This article will analyze and elaborate specifically from two parts: Internet enterprises and traditional industries (banking, insurance and other industries). I hope it will be helpful to you. ?

The mainstream of Internet companies is MySQL?

Open source database MySQL, because of its own characteristics and Internet usage scenarios, is very popular in the Internet community and has a wide range of applications.

Internet companies often have high concurrency, large amount of data and other business characteristics, at the same time, in order to take the lead in the fierce competition, products will continue to iterate, need to constantly launch new products, and do a lot of promotion and operation activities, which can not be predicted in advance from a technical point of view, can only rely on the scalability of IT, so Internet companies have a persistent pursuit of the scalability of the system.

The reason why MySQL is so popular in the Internet community can be summarized as follows:

Although MySQL is open source software, it is easy to use, has high stability, perfect functions, and can be customized according to business characteristics of the required storage engine, performance optimization, so as to adapt to its own business. Many commercial database software design goals do not take into account the high concurrency and mass user scenarios that Internet services often face, which can not meet the basic needs of enterprises.

Because the MySQL code is completely open source, when there are any problems in the enterprise's business, it can be checked and responded as soon as possible to ensure the user experience. However, the core technology of commercial database software can not be mastered deeply, so it is difficult to have the ability to solve problems quickly enough.

Running MySQL on a standard X86 server can greatly reduce hardware costs and save a lot of License costs.

Although MySQL has various advantages, MySQL is not omnipotent. A complex SQL or large table Join may overload MySQL and deplete resources. At the same time, MySQL itself has a serious disadvantage: it does not have a mature high-availability and distributed solution.

Therefore, the choice of most Internet companies is to use mixed use, using MySQL when MySQL can solve problems, while some businesses with higher requirements for performance, security and reliability use commercial database software, and then see if there is a chance to replace it.

Financial industry

The data storage layer of most systems in the financial industry adopts the implementation method of "minicomputer + commercial database + high-end storage array". With the development of business and technology, some problems are gradually exposed.

Security controllable demand, regulators from the height of national information security of the banking IT infrastructure put forward open-source, domestic, security and controllable requirements.

The banking industry is facing increasingly severe pressure of IT cost control. Based on the implementation of the current data storage layer, the data storage cost of each system is millions.

With the innovation and expansion of electronic banking and online banking, the data storage layer lacks good scalability, which makes it difficult to cope with the high concurrency of data access in the application layer.

In the past, banks used high-end equipment, such as minicomputers and large storage, to ensure the availability of databases. In terms of scalability, the processing capacity is improved mainly by adding CPU, memory, disk and so on. With this centralized architecture, the database has gradually become the bottleneck of the whole system, which can not adapt to the huge demand for computing power of massive data.

The financial industry is generally faced with new technical and business challenges brought by Internet finance. Highly available, highly reliable and scalable big data platform and distributed database solution is a new technological choice for the financial industry. It is not only conducive to the financial industry to enhance business innovation and user experience, but also enhance its own technology reserves to meet the market challenges of the Internet era.

Therefore, for the banking industry, the solution of "distributed database + Hadoop big data platform" to gradually replace the existing relational database has become the best choice.

A new generation of distributed relational database arises at the historic moment

Distributed database is one of the important technology selection directions for these large enterprise users (such as e-commerce, finance, manufacturing, retail, etc.) to carry core business, and it is an important technology platform to help enterprises deal with large-scale structured data. In order to meet the actual needs of users for distributed databases and to help traditional enterprises gradually migrate their core business to the cloud, Qingyun QingCloud independently developed a new generation of distributed databases.

Qingyun's distributed Relational Database-RadonDB

Qingyun QingCloud RadonDB is a new generation of distributed relational database developed based on MySQL, which can be expanded infinitely, supports distributed transactions, has strong consistency of financial-level data, and meets the stringent requirements of enterprise core databases for large capacity, high concurrency, high reliability and high availability.

RadonDB architecture diagram

As shown in the figure above, RadonDB adopts a highly available distributed architecture of distributed SQL nodes and distributed storage nodes. The architecture of one master and multiple slaves is used in each partition, and multiple copies of data are stored, which can automatically switch over in seconds and take effect instantly. At the same time, it supports cross-data center deployment and fully ensures the high availability of services.

The storage layer consists of multiple Node, each Node is responsible for part of the data storage, while in the storage node, through the GTID + Raft + Semi-Sync-Replication mechanism to ensure a high degree of consistency of data writing. The underlying hardware generally uses a low-cost X86 architecture storage server.

At the same time, the storage layer uses one-master and multi-slave MySQL as the storage engine, which is different from other distributed databases in the industry (Google Spanner). The classic MySQL is chosen as the storage engine for the following reasons:

MySQL is widely used, and its reliability and stability have been verified for a long time.

Users' MySQL database can be migrated to RadonDB without much modification.

With the continuous evolution of MySQL, its functions are becoming more and more perfect, such as supporting computing push-down, data nearest computing principle, multi-index write atomic guarantee, minimizing data transmission between SQL and Storage layer, and so on.

As shown in the figure above, the data in the distributed database system is interrelated. Although each small table (child table) is scattered, it is logically a unified whole. For the upper application, it can be regarded as a centralized database system.

At the same time, the small table (sub-table) can drift dynamically, and dynamically expand and scale with the heat and size of the table to ensure the optimization of resource allocation. Supports unlimited horizontal expansion of storage nodes, thus providing dynamic and infinitely scalable storage capacity. Performance increases linearly with node expansion, making it easy to cope with the performance challenges of large capacity and ultra-high concurrent requests.

In addition to the above basic features, RadonDB is highly compatible with MySQL syntax, supports HTAP mixed mode, can be deployed across data centers, supports intelligent automatic table sharing, smooth capacity expansion and automatic operation and maintenance, and has zero business interruption during capacity expansion and failover without human intervention. At the same time, it supports HTAP mixed mode, and provides comprehensive service monitoring, audit logs and security protection measures.

As a large-scale distributed database service based on cloud model, RadonDB has all the flexibility, agility, on-demand and light operation and maintenance characteristics of cloud services.

Write at the end

It is hoped that RadonDB database can bring more possibilities to practitioners in the industry, including the new business of the financial industry, the construction of new IT, and the demands of high reliability and high performance of the core business of Internet companies.

In the future, RadonDB will be all open source. I hope more partners can join us and bring more surprises to the industry. If you have more questions about distributed databases, we will have an online launch on December 12, and RadonDB's R & D engineers will talk to you face-to-face, don't miss it.

Qingyun QingCloud next-generation enterprise cloud architecture-- "full model cloud" online launch, look forward to your application.

Http://t.cn/RYAz05V

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.