Architecture Upgrade! GBase Powers Bank of Hebei's Lakehouse Data Platform

In recent years, with the rapid development of technologies such as 5G, big data, artificial intelligence, and the Internet of Things, all types of data have shown exponential growth in scale and diversity, requiring enterprises to process massive amounts of structured, semi-structured, and unstructured data simultaneously. Compared with other industries, commercial banks in the intelligent era rely more on data elements to support their business, and their business processes urgently need richer, more accurate, and more efficient data services to drive business innovation by mining data value.

Facing increasingly strong digital transformation needs, Bank of Hebei, in its data application construction, combined its existing data lake, introduced GBase 8a MPP database to upgrade and migrate the data platform, forming a lakehouse data service system with an MPP + Hadoop technology stack.

Challenges Faced by the Traditional Data Platform

Bank of Hebei's original data platform was built on a Netezza appliance and served as an important foundational platform system within the bank. In 2019, IBM announced the discontinuation of all support for this appliance, while the bank's existing data platform also faced many issues and challenges, primarily including:

Insufficient Data Integrity

The original data platform had insufficient storage space, making it impossible to preserve key analysis results and integrated data for long periods, failing to meet data services with longer lifecycles. Moreover, the data platform and big data platform, as the bank's foundational data platforms, operated independently and lacked integration, so the existing data architecture and technical system could not collect diverse data, resulting in insufficient capability for consolidating data across the entire bank.

Low Data Standardization

In terms of data standards, source systems had inconsistent data standards, and the data platform lacked indirect enforcement of data standards, resulting in failure to form visible, easy-to-use, and effective data assets. Regarding data models, there was a lack of effective management for data models, leading to non-standard data development, inefficient data usage, and insufficient capability for turning data into assets.

Weak Data Timeliness

Data platform resource utilization hit capacity limits and could not be expanded, resulting in insufficient computing power and the inability to complete analysis tasks on time. Processing time was delayed by 7 hours compared to the initial launch, severely affecting the timeliness of critical reports. In addition, the lack of supporting tools led to long data development cycles, further reducing delivery efficiency.

Considering regulatory requirements, Bank of Hebei comprehensively evaluated the need for data capability building in its digital transformation and chose to introduce GBase 8a MPP database to build a new-generation lakehouse data platform, addressing the unsustainable technical support for the original data warehouse and the deficiencies in data capabilities.

Building a High-Performance, Scalable Data Platform with GBase 8a

The bank's data platform is primarily used to store data from various business systems, such as the core system, personal loan system, online banking system, and retail system. The new-generation data platform upgraded from the Netezza appliance to a horizontally scalable distributed architecture capable of supporting massive and complex business data, adopting the fully self-developed China-originated distributed logical data warehouse GBase 8a to accommodate the continuous growth in data volume and complexity of data types across business systems.

Overall Lakehouse Data Platform Architecture Diagram

The entire data platform is built on 16 physical servers. GBase 8a handles all structured data storage and computing tasks, achieving seamless integration with data extraction tools and ensuring full functional replacement of the original system. Meanwhile, in collaboration with upper-layer application vendors, the business migration tasks were successfully completed. GBase 8a processes and transforms data, providing reliable and stable support for generating high-value data. Leveraging its built-in capabilities, GBase 8a seamlessly connects with the existing Hadoop system, supporting the implementation of data models and unified data integration across the bank, forming data organization and models for various data layers. The new-generation data platform is effectively integrated with the data lake, implementing a rational data layering based on the two data ecosystems, forming a complete data lifecycle management system covering collection, management, storage, and usage.

Project Outcomes

The new-generation lakehouse technology platform uses GBase 8a distributed database as the computing engine and the data lake as the primary storage, supporting elastic scaling of platform resources and enabling rapid ingestion and storage of large-scale, multi-type data, thus creating a complete, enterprise-wide data resource. Combining the advantages of both data lake and warehouse, it integrates massive storage with high-performance computing capabilities to meet the bank's multi-level demands for agile data access at a lower cost. The new platform rebuilds data models in a way that better suits business users' habits, refines common data requirements across business lines, and establishes a one-stop data development and delivery system. Additionally, the new system reconstructs the standard data framework, establishing systematic norms for data architecture, model design, and data development, forming a long-term mechanism for data standard management and data quality control, and completing an effective closed-loop for data lifecycle management and governance.

Conclusion

The successful deployment of the lakehouse data platform built on GBase database at Bank of Hebei not only resolved issues in data integrity, data standardization, and data timeliness, but also significantly improved the level of business intelligence through the full integration of big data and MPP technologies, providing a key driving force for the bank's digital transformation. It also laid a solid data and technology foundation for the subsequent construction of management application systems.