Xinjiang Mobile Cloud Business Analysis Project——Low-Cost Cloud Transformation and Upgrade

The China Mobile Xinjiang business analysis system logically consists of an acquisition layer, a data layer, and an application layer. The data layer is further divided into a core database, an information database, and a tag database. The tag database is a data mart application designed for specific marketing analysis, supporting customer analysis, metric statistics, and other tasks to provide foundational data for marketing management. Currently, the tag database runs on DB2, which is experiencing system resource strain and performance bottlenecks during data queries and exports, necessitating an urgent upgrade.

The existing system currently performs 100 to 200 data query and export operations each day. The upgraded system must deliver over 10x performance improvement for these operations, robust scalability, multi-dimensional complex analysis on massive data, rapid identification of target customers, and support for evolving application needs over the next 5 to 10 years.

This project deployed a GBase 8a MPP Cluster data warehouse to build a business intelligence system cluster, consisting of 14 cluster nodes and 2 loader nodes. The current total data volume is 10 TB, with a daily increment of 50–80 GB. Leveraging GBase 8a MPP Cluster's advanced columnar storage, smart indexing, and high-efficiency compression, the system supports storage and high-performance queries of 4 months of daily data and 13 months of monthly data, significantly boosting query and export performance. Cluster nodes are interconnected via 10 Gigabit Ethernet switches, ensuring fast data loading and efficient data exchange. The SafeGroup high-availability mechanism ensures cluster stability.

• Marketing Acceleration: Self-service queries shorten the demand turnaround time, enabling frontline business staff to pinpoint target customers for new services and improving marketing efficiency.

• Cloud Transformation: Move away from traditional IOE (IBM, Oracle, EMC) architecture to reduce scaling costs and achieve linear scalability on demand.

• Performance Boost: Loading and query performance improved by over 10x, supporting concurrent loading and querying to accelerate business responsiveness.