On Friday (July 12) by the giant sequoias database to get involved in the App sponsored Balizhuangyang technology salon shared theme is about distributed database architecture and combat.
The following is the Share be based on the content of the site to share the database giant fir Record finishing.
Introduction to Database Sequoia
redwood trees, next-generation distributed database technology focused research and development, since its inception in 2011, adhere to start from scratch to build a distributed open source database engine, is China's first two consecutive years, Gartner database reporting database vendors.
The main products include Sequoia database SequoiaDB distributed relational database and SequoiaCM enterprise content management software, distributed application scenarios including online transactions, data sets, distributed content management.
Currently, the database giant fir lines on nearly one hundred large commercial bank's core production business, and is widely used in finance, telecommunications, government, internet, transportation and other areas, the total number of enterprise users over 1000.
How Database Development Trend Micro service should respond to
internal enterprise application development are many from the "chimney" development of traditional middleware plus the database, the transition to the micro-service architecture. In the micro-service architecture, almost every micro-services requires the ability to provide data persistence, and users expect the amount of data carried by each micro-services capable of infinite elastic expansion. However, in the process of micro-service architecture, each micro service uses its own separate database to store data will make the past concentrated in one place dispersed to many different devices, the resulting data of the entire IT infrastructure of the heavily fragmented. 
In fact, when the micro-business users of the service architecture, from the perspective of data management, the industry there are two approaches.
The first: that of micro application service reform, the underlying database using traditional centralized database for storage. This practice learning costs are lower, the existence of tightly coupled data, not elastic expansion, and there may be a single point of failure problem.
The second: each group corresponds to a separate micro-service small database, often using MySQL or PostgreSQL, more used in the industry. This mechanism can solve the problem of centralized storage, but it also brings new challenges, including extreme fragmentation of data can not be shared between the micro service, operation and maintenance costs extremely high.
Both approaches can not solve the problem of data storage management in micro-services, distributed databases is to solve these two problems. The first is to do data for each micro-elastic expansion service, the second is to achieve unified management of data throughout the enterprise IT storage in order to avoid fragmentation.
Online trading and what kind of distributed database
online transaction database characteristic
for micro-distributed database service should have characteristics, should be primarily from two dimensions. One is compatible with conventional technology, and second, innovative technology and architecture.
Compatible with the traditional aspects of technology, we must support SQL and ACID integrity.
Prospective new technical point of view, first of all, the core values of distributed database that database resource pool to ensure 100% compatibility with traditional database based on the distributed elastic expansion must be met, when there is insufficient resource pool inside space and computing power, required for expansion is calculated by dynamically increasing the storage node.
Secondly, we adopt the development process, SQL standards, and security policies vary, so the distributed database access interface must be able to support multiple modes.
Finally, HTAP, namely transaction analysis mixing capabilities. Online transaction data and real-time data analysis resource isolation within a resource pool, with access to a database and can be done without disturbing each other. Suitable for micro-services database must have a strong mix of transactional analysis capabilities.
Distributed database architecture and key features of
giant sequoias database instance architecture
to build a database for micro-services architecture, redwood database using a separate computing storage architecture. Wherein the storage layer using self-developed native distributed database engine, the upper layer can be calculated to create hundreds of database instances, with each instance of the application database is completely transparent, without perception. As shown in FIG.
Computing and storage separation architecture
for computing and storage separation, the distributed storage layer to expand, SeuqoiaDB distributed storage engine has many nodes, and which node and the coordinator node catalog data node and catalog the node. Coordinating node is routing of data stored in the data node. Save the entire data catalog node cluster system information. Data node to break up data into different partitions, using the three-copy architecture, any node failure does not affect the normal operation. 
Sequoia core database application scenarios
positioned under the redwood trees large database dimension is a real financial level distributed relational database. Sequoia database currently used in distributed enterprise scenarios including online transactions, data sets and distributed content management.
Online trading is one of the most widely used database scene, often used to support core business operations. Distributed online transaction database includes core business value, distributed architecture transition, high concurrency, high processing capacity, business continuity, scalability and self-control and data security requirements.
Data sets to provide real-time online service the whole amount of data, it refers to all external services outside the traditional core business transactions.
Content management platform to provide enterprises with storage, management and use of vast amounts of unstructured data capabilities. Common applications include imaging platform, document management platform, dual audio and video recording systems.
Distributed transaction capability
Sequoia database supports distributed transactions using the two-stage submission to ensure consistency of data across multiple nodes, the lock mechanism is pessimistic locking, supports read and write of the MVCC submitted ability.
HTAP separate read and write
Multi-tenant physical isolation
By distributed architecture database engine, Multimodel multi-mode data storage types and instances of a database instance management, database giant sequoias can achieve the ability to manage multi-tenant ah, HTAP etc cloud of database necessary technical capabilities.
Finally, the Chairman of the Technical Committee Balizhuangyang Dan and giant sequoias pictures photo database technology experts, expect more database technology exchange.