Provide nationalized analytical databases and solutions to bring 40% computing speed to enterprise databases.
Analytical Processing Database is mainly used for storing and analyzing massive data. At present, the traditional analytical database is represented by Teradata (USA), Oracle (USA), and Greenplum (USA), and has a market share of more than 70% in China for a long time. However, with the development of new technologies such as big data and artificial intelligence, enterprises have to face explosive data volume and more complex processing requirements, while traditional analytical databases have limitations in cost and processing power. Based on the pain point of the market, Juyun Zhizhi has created a purely localized analytical database, which not only has the functions of the traditional analytical database, but also improves the data processing capability and greatly reduces the unit data processing cost. The need for data processing in the era of artificial intelligence.
Juyun Weizhi was established in May 2013. It is an enterprise data warehouse technology service provider, providing analytical database (Linkoop DB) and solutions for military, transportation, telecommunications, finance and other fields.
LINKOOP DB Product Architecture
In terms of data throughput, Linkoop DB (hereinafter referred to as LDB) is easier to implement PB-level bulk data processing capabilities than traditional analytical databases (Oracle, Teradata, Greenplum); and LDB has highly scalable features. The horizontal scalability of data distributed storage and computing can be guaranteed, with more than ten times the data access performance improvement compared to mainstream open source solutions.
LDB has outstanding advantages in data processing timeliness and ease of use. This not only helps companies achieve more and more complex data processing, but also reduces application development thresholds. Strong>
The traditional analytical database can only handle the delay processing of batch data due to the limitation of cost and technical architecture. Typical scenarios such as T+1 report application, and LDB adopts batch integration Optimization technology, not only improve the bulk data processing capacity by 40%Can (TPC-DS standard test), also realizes real-time data processing speed of millisecond level, which can be used to support analysis applications of real-time business scenarios, such as online data analysis and real-time event detection.
In terms of ease of use, LDB uses the standard SQL language and extends the syntax of stream data processing (SQL Stream) and AI computing (SQL AI). It is currently the only batch data processing and stream in China. Data processing and AI computing provide an analytical database with a unified SQL interface. Therefore, the product can achieve a single SQL interface to support mixed applications, without changing the way upstream and downstream developers use, making the code more readable and application development faster.
In addition, LDB has added data processing scenarios for artificial intelligence applications. Dr. Chen Guanhua, the company’s chief scientist, said that the domestic artificial intelligence industry is still in its infancy, and the data management methods of many artificial intelligence applications are still in their original state, which greatly affects the landing process of enterprise-level applications. LDB has unified management capabilities for structured and unstructured data, as well as a distributed execution engine for AI computing. Batch, stream and AI tasks and hybrid tasks are optimally optimized through multi-layer scheduling and multi-stage incremental optimization techniques, improving resource utilization efficiency. For example, the company completed the artificial intelligence landing project of a large international airport, and integrated the historical operation data of the airport and the real-time data of each business system and carried out data modeling to predict the possibility of large-scale flight delay in the future and estimate the flight delay. According to the impact of the overall operational efficiency and security resources of the airport, the matching emergency response plan is recommended accordingly, so that the large-area flight delay prediction can reach the second-order response, and the prediction accuracy rate is over 90%.
The company provides a versatile database product and works with upstream and downstream companies to customize solutions for customers. At present, the company has obtained 50-60 customers mainly through direct sales and channels. The customer price is 1.5-200 million yuan, and the license + service fee is charged. The company will focus on developing the sales market in the future. Dr. Chen Guanhua, chief scientist of the company, said that the effect of data processing depends on three points: data quality, models and algorithms used, and computing power. The first two points require the company’s own accumulation of data and business knowledge. Based on this, the data management and computing capabilities provided by the database enable efficient data processing and high-value data applications. Therefore, cooperation between industries is crucial.
The founding team of the company is from Peking University. The team has more than 10 years of experience in database research and development in head offices such as Microsoft and Teradata. In the early stage, the company focused on technology accumulation, and it was only in the past three years that it began to launch products, and in recent days, Qiantang Star·2019 (No.Winner of the 4th) Entrepreneurship Innovation Competition.