Hangzhou Traffic Management Command Platform Phase II Project — Low-Cost Data Warehouse for Massive Data

lMassive Data Storage: Supports petabyte‑scale data storage and processing, breaks down data silos and integrates information across law enforcement agencies, providing a solid foundation for upper‑layer applications and data mining.

lHigh Performance: Eliminates the cost, processing latency and reporting delays of traditional relational databases, boosting large‑scale transaction processing performance by up to 10×.

lInformed Decision‑Making: Provides objective data for evidence‑based decisions, ensuring real‑time awareness of dynamic information.

lLow Cost & High Scalability: Achieve up to 20:1 storage compression on commodity x86 servers, dramatically lowering hardware investment. The shared‑nothing, flat architecture scales linearly online, resolving the expansion limits of legacy database architectures.

The physical layer utilizes X86 PC servers, while the software layer deploys GBase 8a MPP Cluster to build a big data platform for traffic management and command. The entire cluster consists of 6 data nodes and 1 loading node. Every two nodes form a SafeGroup, totaling 3 SafeGroups.

The Hangzhou Traffic Management Bureau needed to aggregate and analyze all major business data since 2010, including incident reports, intelligence, traffic violations, vehicle passage records at checkpoints, floating car data, microwave sensor data, SCATS (traffic signal controller) data, road segment status, dynamic map coordinates, and derived data based on these sources.

Subsequent data is loaded into GBase 8a MPP Cluster from other systems via database links or ETL tools. Up to dozens of ETL jobs can run concurrently, primarily scheduled by time.

lData Source Diversity: Data needs to be extracted and loaded from over a dozen existing systems within the Hangzhou Traffic Management Bureau, resulting in severe data heterogeneity.

lUnification: Due to performance limitations of the existing Oracle-based transactional database, many analytical workloads cannot be completed. Moreover, data is siloed, cross-system information sharing and correlation are insufficient, and in-depth mining and analysis are lacking. A data warehouse platform that supports massive data volumes, unified management, and unified scheduling must be built.

lCompatibility: Within the established operational environment, intelligent decision support and analysis capabilities must be delivered using big data technologies, enabling complex analytical computations and meeting the demands of data analysis applications.

lForward-Looking: Accommodate the big data platform roadmap for the next five years, with the ability to scale continuously.

With the development of transportation in Hangzhou, the number of motor vehicles has soared and traffic flow has continued to climb. Traffic-related information—such as police incidents, vehicle volume, road network status, violation records, and travel records—has also grown explosively. The traffic police department faces ever-increasing data volumes, resulting in a massive total data pool. Efficiently analyzing this data to extract hidden value, so as to improve infrastructure utilization, elevate traffic management, and enhance traffic safety, has become a pressing challenge.