“Behind technological innovation lies a grand historical background. The booming digital economy in China brings historical opportunities for independent innovation, with databases serving as the foundation and cornerstone of digital economic development. Yunhe Enmo seizes industrial opportunities to realize Chinese creation.”
——Gai Guoqiang, Founder and General Manager of Yunhe Enmo, Kunpeng MVP
The changes in the international and domestic environment make independent innovation in China’s information industry more meaningful in this era, with the localization of basic software being a key focus, and databases being the top priority.
The database field is typified by the openGauss open-source community, which has created a high-performance, high-security, high-reliability, and high-intelligence open-source database product and a thriving ecosystem.
Recently, the openGauss Summit 2023 was successfully held in Beijing. At the sub-forum on database technology innovation and application practice, themed “Harvesting the Fields, Creating Change for the Future,” Yunhe Enmo shared their thoughts and practices regarding the database industry.
Continuous Technological Innovation to Create Products That Users Truly Utilize
As one of the first council members of the openGauss community, Yunhe Enmo is renowned for its deep accumulation in the commercial database field. By refining its technological reserves and abstracting cutting-edge insights in the database field, Yunhe Enmo is committed to “bringing the capabilities of commercial databases into the community,” and the crystallization of this effort is the MogDB database.
MogDB is an enterprise-level relational database launched by Yunhe Enmo, enhanced based on the openGauss open-source kernel, described as “stable and easy to use.”
With more than three years of open-source history and over 1000 days and nights, MogDB has achieved over 1001 excellent qualities, including more than 1000 feature enhancements and a complete machine solution.
-
Over 1000 feature enhancements include compatibility improvements, kernel feature enhancements, reliability enhancements, etc.;
-
A complete machine solution that provides users with an integrated extreme performance and ease of use application experience through separation of storage and computing, multi-node concurrent writing, combined with Yunhe Enmo’s advantageous products zData X and zCloud.
In line with the digital transformation and upgrading pace of industry clients, to ensure that MogDB is truly utilized by users, Yunhe Enmo has equipped a complete toolchain around practical application scenarios, facilitating new system deployment or domestic substitution.
Zhang Wanchuan, General Manager of the Database Product Division of Yunhe Enmo
Zhang Wanchuan, General Partner of Yunhe Enmo’s Original Data Technology and General Manager of the Database Product Division, introduced during his speech that in replacement scenarios, MogDB provides a complete migration toolchain, including compatibility assessment, online and offline migration, and final data consistency verification, forming robust support covering the entire lifecycle of database migration. After migration, MogDB offers suites including high-availability components, installation and deployment tools, and monitoring graphical tools to help users achieve secure and convenient database management.
To simplify the extensive transformation work arising from different application scenarios, Yunhe Enmo has launched diversified product forms and deployment models tailored to the characteristics of different application scenarios and data types. For instance, for clients’ private cloud containerization scenarios, Yunhe Enmo has introduced a containerized deployment product form; for IoT scenarios, Yunhe Enmo has expanded the openGauss kernel to develop a time-series engine and hyper-converged time-series database Uqbar, creating a one-stop IoT data processing platform.
Zhang Wanchuan stated that after more than three years of research and development, MogDB not only inherits the “four highs” (high performance, high availability, high security, high intelligence) characteristics of openGauss but also addresses customer pain points and combines its understanding of mature commercial databases with technical insights, launching a large number of innovative features around core product value points such as extreme high availability, high performance density, enhanced compatibility, and improved usability, while actively giving back to the community.
MogDB 6.0 Will Officially Launch Mid-Year Zhang Wanchuan introduced that the new version will provide brand new migration tools, significantly improving efficiency — the peak value of full data migration can reach 600GB per hour without downtime; incremental migration achieves thirty thousand records per second; by adopting more efficient verification algorithms, the verification efficiency can reach 500GB per hour, making the entire migration process shorter and smoother while also supporting breakpoint resume capability.
MogDB 6.0 will further improve heterogeneous database migration with long-cycle parallel dual-track, supporting bidirectional structure and data synchronization between various heterogeneous databases such as Oracle, MySQL, and MogDB. During the business cutover period and after the cutover, dual-track parallel capabilities will provide safer heterogeneous migration and greater peace of mind for users.
Oracle Compatibility will be significantly enhanced in MogDB 6.0, forming a complete end-to-end capability from upper-layer client tools, driver protocols to the underlying kernel, specifically manifested in:
-
Client tool level, supporting SQLPlus, SQLLoader, Pro*C;
-
Driver protocol level, greatly enhancing compatibility for Python, ODBC, JDBC, and SQL API;
-
Syntax interface level, achieving a large number of syntax adaptations and functional compatibility including PLSQL;
-
Kernel level, achieving DDL and DML behavior compatibility, with DDL automatically committed and DML behavior consistent with Oracle.
“From visible syntax interfaces to invisible behavior mechanisms, MogDB is committed to fully ensuring smooth migration and correct operation of business.” Zhang Wanchuan emphasized.
In query optimization, the newly added scan pre-read capability in MogDB 6.0 eliminates IO latency, achieving scanning efficiency equivalent to full cache scanning, resulting in a maximum 2x improvement; hash operator optimization utilizes CPU cache prefetch to achieve 3x acceleration, and hash join end-to-end improves by 2x, with efficiency comparable to large foreign commercial databases.
In architectural upgrades, MogDB 6.0 will introduce a resource pooling/storage-computing separation A-A (multi-active) architecture. This architecture will greatly enhance product competitiveness in terms of extreme high availability (RTO<10s), high performance density (100T level OLTP core systems), and system elastic scaling, combined with the aforementioned compatibility capability upgrades and query optimization enhancements, enabling MogDB to match foreign commercial databases in large core system scenarios.
Currently, MogDB is accelerating its implementation and popularization across various industries. For instance, Sichuan Tianfu Bank has deployed the MogDB Container (MogDB container version) cluster on its existing container platform to support current business, while combining the MogDB cluster unified operation and maintenance management platform MogDB Stack to form management and monitoring capabilities covering the entire lifecycle of the database; based on this, it connects with upper-layer applications to construct a self-innovative “full-stack containerized database management new model,” that is: container platform + MogDB container + MogDB Stack + APP container, achieving unified scheduling, allocation, and management of full-stack resources.
Whether in functional updates or industry practices, MogDB’s continuous iteration in compatibility and performance optimization demonstrates the vigorous vitality of Chinese database technology in continuous breakthroughs.
Zhang Wanchuan concluded that the key to driving innovation is finding and utilizing variable factors, such as hardware updates, algorithm upgrades, resource model changes, and user application demand expansion. “Hardware innovation drives algorithm upgrades such as vectorized queries and lock-free concurrency to improve resource utilization, resource clouding gives rise to cloud-native elastic architectures, and the diversification of large models and user applications promotes the development of multi-modal databases and the expansion of relational database capabilities.” MogDB’s path of database innovation is full of expectations.
Looking Seriously at the Future, Yunhe Enmo Has Taken a Unique Path
Turing Award winner James Gray once said, “If you have vision, you must pursue visionary research. Spend as much time as possible doing things you are proud of and try not to do meaningless things.”
How far into the future can we see if we focus on the future? Gai Guoqiang, Founder and General Manager of Yunhe Enmo, said that when we look seriously at the future, we can indeed see far.
Gai Guoqiang, Founder and General Manager of Yunhe Enmo
Yunhe Enmo is deeply involved in the openGauss ecosystem, integrating academic and industrial accumulation, and delving into the database field to do things they are proud of.
For example, Yunhe Enmo’s intelligent algorithm team is exploring intelligent-related initiatives, training DB-GPT based on large models, achieving intelligent Q&A based on knowledge base systems, and interfacing with external applications through open APIs to meet database operation and maintenance, DBA diagnostics, and other work needs. The DB-GPT practice based on the MogDB knowledge base can answer specific questions about database characteristics without the need to flip through documents.
Gai Guoqiang stated that when it comes to AI and large models, intelligence and databases, we need to find a balance between precision and ambiguity. Large model intelligent agents (LLM Agents) can continuously help improve Q&A accuracy. For instance, providing precise evidence is an effective method; secondly, introducing external knowledge bases for information supplementation or confirmation is also a feasible way.
Traditionally, data is centered around storage rather than use. The storage model determines how data is used. When models change, the significant differences between models may lead to the need to start over with previous work. So is there a better way to allow developers to declare the intent of data usage rather than the format of data storage, and then let the database generate matching data formats and APIs based on application scenarios?
The Data Intent Language (DIL) has emerged to describe the purpose and format of data, which can transform data application patterns. SQL domains are dictionary objects belonging to schema that encapsulate a set of optional properties and constraints for common values. SQL domains provide constraints, display, sorting, and annotation properties. After defining an SQL domain, you can define table columns associated with that domain, thereby explicitly applying the optional properties and constraints of that domain to those columns. Therefore, SQL domains are used to provide additional information to stored columns (JSON or relational) and are thus used to define and validate data.
Inspired by this, Yunhe Enmo has enhanced aspects such as database observability, automated operations, and intelligent operations, turning its insights and understanding of the industry, along with user demands, into product capabilities. For instance, MogDB has built-in time models that support load models for active session outputs; the MogDB operational monitoring visualization, based on zCloud, supports real-time alerts, elevating observability to a level comparable to large foreign commercial data; in terms of autonomous intelligence, intelligent diagnosis based on fault trees and intelligent baselines based on machine learning achieve simplification of operations to DBA-less.
Regarding the company’s development model, Gai Guoqiang mentioned that Yunhe Enmo has taken a unique path. Starting from database management services, expanding to database management software, and then building the core product MogDB, followed by constructing a complete ecological matrix around MogDB.
Currently, Yunhe Enmo is engaging in extensive ecological cooperation with manufacturers of hardware, complete machines, cloud infrastructure, chips, operating systems, middleware, backup storage, and application software. Several of the company’s products have completed compatibility adaptations with partner products, building comprehensive solutions for customers. At the sub-forum site, Yunhe Enmo completed a strategic cooperation signing with Baixin and launched a full-stack domestic database integrated machine solution.
“The future world is an intelligent world, and the driving force of the intelligent world is data. We position ourselves as intelligent data technology providers, helping users realize a data-driven intelligent future. In this vision, Yunhe Enmo is doubling its efforts to gradually bring intelligent data technology into our products, assisting users in digital construction and intelligent evolution,” Gai Guoqiang concluded.
Data-driven, achieving the future, Yunhe Enmo, lives up to expectations!
Founded in 2011, Yunhe Enmo’s mission is “Data-driven, achieving the future,” and it is an intelligent data technology provider. We are committed to bringing data technology to every industry, every organization, and every individual, building a data-driven intelligent future.
Yunhe Enmo provides reliable products, services, and solutions in the fields of data bearing (distributed storage, data continuous protection), management (database infrastructure software, database cloud management platforms, data technology services), processing (application development quality control, data model control, digital transformation consulting), and applications (data service management platforms, data intelligent analysis processing, privacy computing) for various organizations, continuously creating value for customers around user needs, stimulating data potential, and striving for an agile and efficient digital world of the future.
