With the development of cloud era, the story of database continues. Whether it is a commercial Oracle or an open source community version of MySQL, database technology has developed from chimney to Federated, then distributed to distributed, cloud, service cloud to cloud, and cloud to cloud native design on DIY (Do It Yourself). They are getting more and more attention and influence all aspects of modern enterprises with prairie fire. So what are the technical principles and the inside of the database?
As we all know, in the context of cloud computing, enterprise IT business is heading for trans regional and global deployment, and IT application software is gradually becoming cloud and distributed. As the key to carrying out business and application, database is based on Cloud Scene Architecture design, and has the ability of cross regional distributed deployment.
As mentioned above, the evolution of database technology is a process from chimney to Federated and distributed.
Database technology evolution
The chimney type database generally adopts the multi copy deployment architecture, and the unified response to the APP reading and writing requests through Proxy is mainly applicable to the single database system with relatively stable access traffic and low pressure, which supports the single service system. Its main challenges lie in usability and scalability. Availability depends on specific solutions, and generally does not possess the capability of Region disaster recovery. The scalability is limited by the number of extended nodes from the library and the maximum instance performance.
Federated databases deploy data into each data node and respond to APP's reading and writing requests through database middleware. It is mainly applied to loosely coupled and heterogeneous multi database systems, which are used to integrate multiple business system scenarios built vertically. The main challenges are usability, scalability and functionality. Availability is limited to data fragmentation, that is, Sharding scheme. Extensibility needs to be expanded by manual operation sub library, large risk, downtime, no cross database transaction and connection in function, and no support for complex SQL queries.
The SQL layer, executive layer and storage layer of distributed database are decoupled and extended independently. It is mainly suitable for face Xiang Yun distributed scene, and requires extreme scalability and high availability database system to support large scale business systems across the region. As the deployment needs sufficient cross regional data center machine room and computing, storage, network and other resources, management needs the ability to control the database system with a cross regional data center, so the main challenge lies in the deployment and management aspects.
It is worth mentioning that the evolution of database technology is closely related to the change of enterprise application scenarios in the new era. At present, the proportion of cloud on the global enterprise application is currently around 50%, and will increase to about 60% in two years. The acceleration of enterprise application on cloud drives the database to accelerate Xiang Yun migration. The traditional database management system DBMS transfers Xiang Yun quickly, transforming it into a database cloud service DBaaS, which continues to support enterprise applications.
Earlier, the business cloud model was mainly DIY, that is, users need to purchase cloud servers by themselves, and install, deploy, maintain and upgrade the database by themselves. The advantage is that it takes advantage of the convenience and reliability of cloud infrastructure services. The disadvantage is that the operation and maintenance cost and the traditional way of the cloud are not constant. The use of commercial database still pays the expensive license authorization cost, the database specifications change and upgrade difficult, the backup and disaster recovery are complex and difficult to implement.
Later, it evolved into a service-oriented model. Users only need to buy database cloud services directly, without the need to buy a cloud server, and no need to install, deploy, maintain, and upgrade the database. The advantages of the service mode are that the application is not modified, the database is open and used, the demand is paid, the flexibility is expanded, the demand is upgraded on demand, and the reliability guarantee of cross AZ and cross Region is guaranteed. The disadvantage is that the user's use habits need to be changed a little.
In the era of cloud original design mode, users will choose a new distributed database service based on Cloud Architecture Design. The advantage is that the database is open and ready to use, pay by demand, and can be deployed over large scale, super resilient expansion, upgrades on demand, reliability guarantee across AZ and cross Region. The disadvantage is the cost and learning cost of the application with cross region reconfiguration, and the user has a certain software design and realization ability.
Cloud pattern on the database
It is not hard to see that enterprises have a long way to go to realize the cloud on the database through external force. However, in the ever-changing market and the fleeting opportunities, how much time are left for the enterprises?
The Ninth China database technology conference is open at the time. CalvinSun, the chief architect of HUAWEI cloud database, will attend the conference and bring two keynote speeches of "HUAWEI Cloud Native distributed database technology" and "MySQL cloud database performance optimization and development trend".
Calvin Sun joined the HUAWEI Institute in Toronto, Canada, in 2017, and made a special presentation at a number of database conferences, with more than 20 years of database kernel development experience. He had been a senior consultant on the Oracle cloud service team MySQL cloud services; he was the head of the Twitter MySQL kernel team; also the head of the OracleInnoDB development team, during which the team led the development of InnoDB MySQL 5.5 and 5.6. Earlier, he served as the head of the MySQL storage engine development team.
In this lecture, Calvin will introduce the key features of the HUAWEI Cloud Native distributed database, such as high reliability, high performance, and easy to expand, and focus on its technical principles and unraveling the technology behind it. In addition, we will review the key performance features of MySQL, introduce the status of MySQL cloud services, and the work of the HUAWEI cloud to improve the performance of MySQL on the cloud, and discuss the challenges and opportunities to further improve the performance of the cloud on the cloud.