On August 26, the Global Distributed Cloud Conference was held in Beijing, which was dedicated to promoting the development of distributed cloud computing in China and playing a leading role as a demonstration for the deepening of digital transformation of enterprises, the development of smart cities and the integration and application of emerging technologies.

　　The conference focused on the stage development results of distributed cloud computing, and AntDB contributed to the commercial promotion and popularization of domestic distributed database by virtue of its years of R&D accumulation in the database field and profound practical experience in the communication industry.

Figure 1: Global Distributed Cloud Conference - Beijing Site

　　In the Database Topic Forum, Mr. Zhang Hua, General Manager of AntDB Product Center, shared the keynote speech of "AntDB Distributed Database Evolution and High Availability Solution", and gave a detailed introduction on the challenges faced by the scale application of AntDB distributed database, how to build the core system application architecture, and the difficulties of business full migration.

　　Algorithm requirements: a home-grown database serving 1 billion users

　　The R&D of AntDB started in 2008, witnessing the ultra-high-speed development of domestic communication industry from 2G to 5G. During these 14 years, the scale of telecom subscribers has continued to surge, and the business forms have been pushing out new ones, which have higher and higher requirements on the support system and database bearing. In the past few years, China Mobile has migrated its core billing, CRM and other services to domestic databases, which is one of the key points for the transformation of operators' business support systems.

　　AntDB builds high-availability solutions through enterprise-class enhancement features to assist China Mobile to complete the independent and controllable transformation of domestic databases in application scenarios such as core transactions and big data analysis.

Figure 2: the evolution of AntDB in the development of 2G to 5G in communication industry

　　There are two challenges during the process:

　　· First, the operator's business support system is responsible for the aggregation of business channels, top-up payment and account records of tens of millions of users, and provides terminal services directly to users, which is one of the key points in the transformation of the business support system, and the reliability requirements are extremely high.

　　· Second, the volume of business is very large, more than 100 billion per month, the overall business volume is not inferior to some large Internet companies. With the scale of 5G services, the data processing volume is still rising, so the operator’s business support system has very high performance requirements for the database.

　　As one of the earliest general-purpose enterprise-level database products, AntDB has been ensuring continuous and stable operation of carrier business system and providing online services for more than 1 billion users in 24 provinces across the country with tens of billions of transactions processed daily.

　　AntDB has four salient features:

　　Amount - A huge number of users. More than 1 billion mobile phone owners using AntDB in the background.

　　History - Long time of implementation. AntDB has gone through 14 years of trials and tribulations since its implementation in billing system in 2008.

　　Speed - Fast execution performance. All the call records and every message first sent on the Internet in the telecom business need the support of AntDB in the background, and AntDB can handle millions of core communication transactions per second at its peak.

　　Stability - Stable and reliable products. AntDB has been running smoothly, safely— in the core system of telecom operators for more than ten years.

　　Scenario innovation: full-stack development for future business

　　In the face of the system construction needs of hundreds of millions of users, traditional databases often require multiple sets of technologies to support: a set of relational databases to achieve metadata management and tagging management, and a set to manage unstructured files, with complex technical architecture, leading to difficult project management and high implementation costs. At the same time, the centralized storage architecture cannot be flexibly expanded as needed to meet the needs of business growth and massive data management.

Figure 3: traditional database to distributed database

　　In the case of  the cooperation with China Mobile in one province, AsiaInfo worked with China Mobile and achieved the highest standard of high-availability architecture in the industry: a two-location, three-center architecture, with the entire system consisting of two clusters of one master and one standby plus a disaster recovery cluster in an off-site center, each cluster consisting of 16 nodes, and three clusters with a total of 48 nodes. By using such a database cluster to support one billion or several billion transactions per day and achieve Telecom-level high availability.

　　Zhang Hua said that from the initial transactional system based on high intensity and high density, to open source ecology and the compatibility with foreign databases, to the overseas support for Telnor Group's business system, AntDB has become more and more generalized, standardized and highly available, which can thoroughly solve the needs of transactional scenarios, and also take into account the analytical scenarios.

Figure 4: the application effectiveness of AntDB’s super OLAP capability

　　AntDB is a HTAP type database, which fully proves the functional advantages of database to analyze and compute OLAP scenarios. The self-developed distributed parallel processing specific functions including heterogeneous indexing and multi-table association improve the data processing efficiency in distributed big data scenarios, and the performance is improved by ten times to tens of times, which meets the business needs of customers for second-level response of massive data analysis.

　　Consistent service: support both centralized and distributed deployment

　　AntDB is a one-stop data platform, i.e., it supports centralized deployment and distributed deployment; it unifies multiple types of data, and one set of service can cope with both transactional and analytical application scenarios, with low maintenance cost and high practicality; based on the native distributed design architecture, it realizes online elastic scaling, strong consistent transactions, cross-room high availability and other capabilities, helping core systems in transportation, finance, and power to solve the problem of massive data management.

Figure 5: the architecture of AntDB’s distributed database product

　　AntDB’s native distributed architecture suits the heavy load scenario of ultra-high users volume with better performance in underlying planning and control of data, which ensures the consistency of transactions and the integrity of data; it also has the features like easy employment and flexible architecture, which guarantee database capability of full functions even in standalone mode after quick employment.

　　A department of China Mobile in one province served by AntDB, which provides the whole process management of customer service on the business side, has high requirements for system stability, service continuity and data accuracy. The project built a read-write separation and micro-base architecture design based on AntDB to support data migration in a short period of time with high quality, and after the migration was completed, the average response time of business acceptance was improved by nearly 100 times, and the unified management of various business channels, such as online and offline, own and cooperative, was realized.

　　About AntDB

　　AntDB was founded in 2008. On the core system of operators, AntDB provides online services for more than 1 billion users in 24 provinces across the country. With product features such as high performance, elastic expansion and high reliability, AntDB can process one million core communications transactions per second at peak, ensuring the continuous and stable operation of the system for nearly ten years, and is successfully implemented for commercial purpose in communication, finance, transportation, energy, Internet of Things and other industries.