Recently, the Global Distributed Cloud Conference (Beijing Site) was held, and AntDB was invited to attend the conference. At the conference, the technical leader Bei Mo delivered a speech with the theme of "AntDB hyper-converged streaming real-time data warehouse, ushering in a new era of distributed database".By sharing AntDB's research and development practice in database frontier technology, we exchanged the idea of new era of distributed database development together with the participants.
Trace back to the original intention, to better devote to product development and service
Due to the late start of China's information industry, we have always relied on European and American brand for database technology. As the size of China's population continues to grow and communications needs can not be met, China's data and information processing volume is "rarely massive in the world". With the exponential growth of cell phone users in China within a short period of time around 2000, the traditional database products grew in the European and American markets were not able to meet the data management needs of such a large volume, especially in core application scenarios such as billing and signaling analysis. Against this background, AsiaInfo, as a partner of carriers, took the lead in technological innovation and developed the first generation of AntDB, the in-memory database, to provide effective support for carriers' massive data and high-frequency transaction scenarios.
Figure 1: evolution of AntDB streaming real-time data warehouse
With the iteration of telecom technology from 2G to 5G, comes exponential growth in volume, quantum leap in subscriber numbers and a big explosion of application scenarios. As a partner of telecom carriers, AsiaInfo Technologies can deeply perceive and foresee the changes of scenarios brought by technological progress, and can conduct the evolution and upgrade of product technology and services with future-oriented insight. AntDB has also completed seven major iterations in this process - from in-memory database to highly available full-featured relational database, to full-stack database compatible with MySQL, PostgreSQL and Oracle, and further to distributed, multi-mode cloud-native database, and now to version 7.2. Each step has been at the forefront of technology, helping customers to effectively cope with data management brought about by new technologies.
AntDB has formed four features during its development:
Amount - A huge number of subscribers. More than 1 billion mobile the phone owners using AntDB database in the background.
History - A history of solid implementation. It has been applied in the core billing system of telecomm carriers since 2008, and has experienced 15 years of careful refinement.
Speed - Fast execution performance. AntDB can handle millions of core telecom transactions per second at its peak.
Stability - A stable and reliable product. It has been running stably in the core system of telecom carriers for 15 years.
By integrating and innovating, domestic distributed databases ushered in a new era of development
Innovation and application of information technology are an important part of China's new infrastructure. Under the auspices of the policy of information innovation, AntDB, as a domestic database independently developed by AsiaInfo, takes up the mission and researches the future-oriented database technology, and brings the development of China's database into a new era with hyper-converged streaming real-time data warehouse: in the overall architecture, AntDB puts forward a new "hyper-convergence" concept, which integrates multiple engines and capabilities, realizing the integration of five major capabilities: memory computing, transaction, analysis, stream processing and time-series, so that one product can provide all services to customers; through the new generation of stream real-time data warehouse engine, stream computing and database are integrated and innovated from the kernel level, and stream processing and SQL are organically combined together.
Figure 2: AntDB hyper-converged streaming real-time data warehouse, ushering in a new era of distributed database
In application, AntDB's hyper-converged architecture under the unified framework effectively reduces the complexity of IT architecture and shortens the project development time by nearly 50%; it supports standard SQL statements for business development, and the business response speed can be increased by nearly 90%; the unified technical architecture effectively solves the consistency problem of data caliber and makes database operation and maintenance management simpler and easier to use. AntDB's new generation streaming real-time data warehouse engine integrates real-time streaming processing into the database engine that has not changed for 50 years, and addresses the streaming processing events in real-time business through its own database kernel capability, and carries out real-time pre-processing of the massive business data generated. From passive query to active push, from real-time analysis to output results, it realizes real-time analysis and alerting of core business data, helping customers grasp the operation status in real time, adjust the operation strategy in time, and obtain data value more quickly.
Case showing the business value
At the end of the presentation, Bei Mo demonstrated the real-time service scenario of charging pile, so that the guests could know the business value of hyper-converged streaming real-time data warehouse more directly.
In recent years, increasing people choose electric vehicles as a way to travel, but "mileage anxiety" is plaguing the development of the tram economy. How is the status of charging piles on the road? How many empty spaces are there? There is an urgent need for an intelligent monitoring service system to integrate and share the information of charging piles around the world.
So, what is the role of the streaming real-time data warehouse engine for real-time business? For the real-time service scenario of charging piles in a province, data such as vehicle information, current consumption information, charging pile status updates, charging volume, etc. in each region continue to arrive, and the system needs to process all kinds of new arriving data in a timely manner, and continuously generate and output the current number of empty piles, dynamic monitoring of today's revenue, and summarize the reports of operating status such as revenue of each region, time-differentiated statistics, and overall revenue distribution.
Then, down to a certain area to see the current distribution and operation status of charging piles very clearly.
Figure 3: example of real-time service monitoring of charging piles
In fact, the real-time alerting and BI display system based on traditional relational database is an offline batch processing mode, which can successfully handle massive data at TB\PB level, but cannot cope with the demand of real-time data analysis and real-time response to news events, and cannot meet the actual demand of real-time.
AntDB's streaming processing mode emphasizes the speed of data processing, and the streaming real-time data warehouse engine can process the newly arrived data in time, so that decision makers can see the latest data changes, accurately grasp the development trend of the situation, and make timely and effective response to the unexpected events.
Top comments (0)