Beijing Housing Provident Fund Management Integrated Information System Upgrade Project

1. Project Background

With the continuous growth in the number of housing provident fund contributors and the ongoing accumulation of existing data, the Beijing Housing Provident Fund Management Center has amassed vast amounts of provident fund data. The traditional data management architecture, built on an Oracle plus minicomputer model, suffers from severe performance bottlenecks amid daily increasing data and restricts scalability. Therefore, the Center's comprehensive information system urgently needs upgrading and transformation to resolve current performance and architecture issues.

2. Business Requirements

To address the performance and scalability limitations of the existing system, this upgrade project requires:

• A distributed database designed for big data to accelerate ad-hoc queries on business data;

• A distributed, open platform architecture that scales seamlessly as data grows;

• Domestically developed products that meet core government business application needs while ensuring security and controllability.

3. Delivery Requirements

The upgrade must meet standards for functional completeness, performance efficiency, security, and terminal compatibility. Upon completion, all business operations will be migrated to the new system and go live.

The Housing Fund Integrated Information System modernization project deploys General Data Technology's GBase 8a MPP Cluster database to build a unified data storage and management platform. It aggregates third-party data from employers, banks, and credit systems, unifies it with the housing fund center's own business data, and performs initial aggregation and rule-based reorganization and computation within the database—such as constructing data cubes and building subject-oriented data marts.

The housing fund system runs on an 8-node GBase 8a MPP Cluster—4 management nodes and 4 data nodes—handling all structured data storage and processing, with business workloads already migrated into production. Two loading servers handle periodic external data intake. Internally, the database is layered into ODS (Operational Data Store), DW (Data Warehouse), and DM (Data Mart) tiers that work in concert to transform raw data into high-value, business-centric information.

Figure 22-1: Architecture of the Housing Fund Integrated Information System

Unlock Data Value: Achieve unified and integrated management of housing provident fund business data, delivering more reliable and timely data to upper-layer applications;

Low Cost and Scalable: Based on an x86 PC server architecture, the platform reduces investment and ensures scalability in line with long-term planning;

High Performance: Data loading, aggregation, and query speeds are 10–20x faster than traditional databases.