news
Recently, the Beijing Fintech Industry Alliance, guided by the People's Bank of China, announced the results of the 2023 Top Ten Projects, and the "Research on Key Business Scenarios of Resource Pooling Database Finance" led by the opengauss community won the awardTop 10 projects of Beijing Fintech Industry Alliance in 2023.
In 2023, OpenGauss released 50.0 and 51.Version 0 realizes the two major architectural innovations of DataPod and DataKit, with hierarchical decoupling and resource pooling, and continues to lead the development direction of database technology. After the launch of the DataPod resource pooling architecture, it has penetrated into the application scenarios of key financial services, and has been supported and praised by customers in the financial industry, and won this award as a technological innovation project, marking that the specific practice of OpenGauss DataPod resource pooling architecture in financial scenarios has been recognized by the industry.
At present, the database transformation of the financial industry is gradually deepening to the core key business, and the challenges are increasing in terms of sustainability, stability, reliability, performance, and efficient use of resources, energy conservation and emission reduction. Resource pooled databases have become an important hot topic in the development of global database technology because they break the traditional architecture and full-stack optimization to multiply the full-stack capacity of databases quickly, efficiently, and at low cost. Led by the OpenGauss community and jointly participated by China Minsheng Bank Co., Ltd., Huaxi Co., Ltd., Industrial Bank Co., Ltd., the project of "Research on Key Business Scenarios of Resource Pooling Database Finance" came into being.
This project has an insight into the needs of banks and key business scenarios, completes the research on the technical principles of database resource pooling architecture in improving reliability, single database performance and optimizing resource use, and tests and verifies it in combination with banking and industry application scenarios, summarizes the value of resource pooling databases, improves the direction, and provides reference guidance for the transformation and replacement of financial core business databases.
According to the results of technical demonstration and verification, the resource pooling architecture has greatly improved the comprehensive capabilities of reliability, performance, and resource utilization, which solves the "impossible triangle" of the traditional architecture to make trade-offs between performance and reliability, or the number of hardware in exchange for performance and reliability, and greatly improves the full-stack capability of the database
One is reliability. The shared storage write-multiple read-a-use architecture ensures no data loss without affecting performance, and the shared disk lock technology shortens the failure recovery time of computing instances by half, prevents split-brain and implements two-node provisioning, reducing the minimum cluster specifications. Through full-stack optimization with enterprise-level all-flash storage, it solves the problems of sub-health of disks, rapid data recovery and reconstruction, and reduces or eliminates the business interruption and performance impact of key component failures and recovery processes. As a result, a reliability system with the core of reducing the impact of failures and redundancy has been built, which meets the requirements of financial services for long-term stable operation.
The second is resource utilization. Thanks to the performance expansion and reliability redundancy, the utilization rate of computing resources in the resource pooled database is increased by 33% and the storage utilization rate is increased by 50%, which can realize the on-demand configuration of computing and storage resources, which is of great significance for financial enterprises to reduce the power supply demand of computer rooms and realize energy conservation and emission reduction of green finance.
The third is performance. In the performance comparison of banking production configurations, the performance of the resource pooled database is improved by 1. compared with the non-resource pooling architecture because it does not need to sacrifice performance to meet the reliability requirements6x, showing excellent balance. In the transactional business simulation test, the resource pooled database improves the performance by 35% and reduces the latency by more than 12% compared with the traditional non-pooled database. In the extreme performance test, the performance of the resource pooled database with full-stack optimization, such as Kunpeng server NUMA affinity high-performance atomic instructions, database user-mode file system, NOF storage network, and all-flash storage, is the same as that of the non-pooled database, indicating that the full-stack optimization of resource pooled database can solve the performance bottleneck of the storage-compute separation architecture and clear the obstacle for architecture optimization.
In summary, resource pooling database is an attempt to jump out of the traditional optimization idea, which not only improves the capacity of a single database database, but also removes the performance and cost barriers of using innovative infrastructure, and creates conditions for further improving the full-stack capability of the database through full-stack optimization.
With the maturity of technology and the enrichment of products, it has a wide range of prospects for business needs such as smooth migration of databases, reduction of database splitting, and support of low latency in financial services.
This project is based on the OpenGauss open source database, which represents the mainstream level of resource pooling databases in China. In the future, the OpenGauss community will further promote the innovation of resource pooling architecture in the context of the digital transformation strategy of the financial industry and the new requirements of database technology financial application specifications.