How a First-Prize Winning Article Was Crafted for the GBase Technology Article Contest
Recently, GBASE and the Modb community jointly held the First GBASE Technical Article Contest. The event received broad support from participants across various industries, with hundreds of article shares and over 6,000 users casting votes, ultimately selecting first, second, and third prizes in the Preferred Articles category. Database engineer Huang Yonghou from the Operations Management Department of a ministry-affiliated unit won the First Prize of the Preferred Articles with his article “Data Platform Embarks on a Transformation Journey: Performance Practices of GBase 8a in Complex Query Scenarios,” which garnered 1,708 votes.
About the Author
Huang Yonghou – Database Engineer, Operations Management Department, a ministry-affiliated unit
He has been engaged in data management for over twenty years, primarily working with Oracle databases. He experienced the frenetic Oracle era under Dongfang Longma (Oracle 7.1.6/7.3.4/8.1.7, etc.) and later led the development of a homegrown database product based on PostgreSQL. He focuses on performance optimization and high availability of database products and has accumulated considerable experience.
Huang Yonghou said: To build sustainable business operations, we plan to introduce an MPP database cluster architecture to handle existing and future massive structured data for complex analytical queries, thereby optimizing resource utilization of the core database platform. The shift to domestic databases has been a technical highlight in recent years. From both technical and policy perspectives, it will significantly influence the sustainable development of our data platform. By conducting POC proof-of-concept testing and building technical reserves, we aim to accelerate the transition to a homegrown database platform and hope this can provide a reference for peers.
Impressions of GBase Database
I began exploring and gaining in-depth understanding of the development of homegrown and open-source database products at the end of 2018. After focusing on GBASE’s GBase 8a, a large-scale distributed parallel database cluster system, I found that GBase 8a is widely used in China’s analytical MPP database market. In May 2021, with a desire for systematic learning, I enrolled in the GBase 8a online training camp. Driven by professional habits and personal interest, I set up a relatively simple environment for prototype verification testing, which helped validate the strengths of different types of database products. For analytical data requirements, I organized this article based on metrics such as data loading efficiency, data type compatibility, structured query friendliness, and complex query performance, which significantly changed my view of the performance of homegrown database products.
Views and Suggestions on the Development of Homegrown Databases
In the long term, replacing with domestic databases is an inevitable outcome. Currently, while ensuring business continuity, we are conducting research, testing, and prototype verification on major homegrown and open-source database products. When conditions permit, we will deploy suitable business applications on homegrown database products to gain migration and operational experience for replacing core systems.
However, for domestic database vendors, users often directly compare their products to established enterprise databases like Oracle, which is somewhat unfair. They thus face the following challenges:
1. Database R&D demands highly skilled talent with broad and deep knowledge and extensive data management experience. Building a database in isolation or purely from theory can hardly create a product that truly meets market needs. The lack of such talent is an urgent issue, requiring significant investment.
2. Database product roadmap planning: Customers often focus on operational friendliness (continuity of user habits), data consistency, zero data loss, high availability, clustering, disaster recovery, backup and recovery, data export/import, and data migration—key aspects of database operations. With the widespread use of the Internet and IoT, enterprise data volumes have multiplied dozens of times compared to 20 years ago; terabyte-scale data is now common. For database products, requirements such as multi-node, cloud-native, and K8s containerization are becoming mainstream trends, leading to the development of auxiliary products like data routing. Currently, most industries’ core business still relies on relational databases. Whether it is traditional single-machine relational databases or the recently emerging distributed relational databases, cloud-native distributed relational databases, and cloud-native HTAP databases, product managers must strike a balance in functionality. The design of a database product must first ensure usability, and then strive for a good experience. Achieving this alone can already be half the success.
The technical barrier for databases is high, but open-source databases like MySQL and PostgreSQL have relatively lower entry thresholds. However, as a DBA with many years of experience, I would like to tell newcomers: the database is usually the single core of the entire technology architecture, with very high availability requirements and complex internal mechanisms. It often involves cross-interactions among underlying storage, networking, operating systems, dynamic business logic, massive data processing, high-performance query response, zero data loss, and disaster recovery. Systematically learning database administration, performance optimization, and internal mechanisms is greatly beneficial for rapid professional growth.