【GBASE Case Study】GBase 8a MPP Assists ABC's Big Data Platform Cluster Relocation Case Analysis

As data volumes at Agricultural Bank of China (ABC) grew steadily, the GBase cluster deployment expanded year over year. The bank now runs more than 1,000 nodes across dozens of cluster environments. To streamline server management and reduce operational costs, ABC’s head office decided to relocate all GBase cluster servers from its Yizhuang facility to the bank’s Daoxiang Lake Data Center on Beiqing Road.

ABC’s enterprise-level big data platform began construction in April 2014. By April 2016, the primary data warehouse was completed as a 121-node active-active cluster. Subsequently, multiple data marts were built around business themes. GBase 8a MPP powered 17 mart clusters, one primary warehouse cluster, and 8 analytical processing clusters. The platform manages more than 20 PB of source data. It consolidates, stores, and governs high-value data across the bank, providing enterprise-wide data sharing services for upper-layer applications. It delivers massive data processing and analytics, along with high-performance, efficient data mining, cleansing, and analysis.

The rapid growth of enterprise data and extended data lifecycles led to insufficient storage and processing headroom. Capacity expansion, relocation, and consolidation became necessary to unify server management and cut ongoing operational costs. For these reasons, ABC decided to move all Yizhuang GBase cluster servers to the Daoxiang Lake Data Center on Beiqing Road.

1. Conduct simulation drills for the active-active cluster switchover in advance to detail the steps, assess potential issues, and implement resolutions.

2. Apply the active-active cluster switchover and relocation mechanism to existing clusters. Relocate Cluster B first, then enable data synchronization between Clusters A and B the following day to ensure B matches A. Relocate Cluster A servers within an appropriate maintenance window.

3. For environments without a Cluster B, temporarily deploy a new Cluster B at the target site. Synchronize data from Cluster A to the temporary Cluster B, and once data is consistent, relocate Cluster A.

4. Ensure GBase technical staff fully support ABC's operations team throughout the move, including post-relocation cluster health checks and troubleshooting.

• Full Inventory of Data Assets
Conduct a comprehensive audit of business data to build a data map, laying a solid foundation for business applications and data access.

• Continuous Data Quality Improvement
Steadily improving data quality is the cornerstone of value.

• Enable Data Interconnectivity
Drive digital transformation by breaking down data silos and connecting business data across the entire organization.

• Accelerate Data Access
Make data readily available whenever needed, drastically reducing data preparation time.

• Ensure Data Security and Compliance
Security is the non-negotiable baseline of data asset management.

• Sustain Data Value Delivery
Continuously and dynamically evolve alongside technological shifts, market changes, and industry advances to unlock recurring value.

Manage, integrate, analyze, and mine the value of big data to help customers achieve risk-controllable, secure, and compliant data assets, improve performance, and consistently deliver innovative big-data services. This is realized through a combination of systems, standards, procedures, management, integration, and analysis practices designed to leverage data assets more effectively and create increased business value.

Services: Data services built on deep, long-accumulated expertise
Products: A rich, full-stack product portfolio
R&D: Strong technical R&D capabilities
Technology: Nearly 20 years of product and technology refinement
Investment: Nearly 2 billion RMB in R&D investment