Fujian Mobile Signaling Monitoring Project——Rapid Query Response to Network Failures

Fujian Mobile’s core network interconnection relies on signaling communication between devices. At the same time, signaling monitoring—covering deep analysis and real-time surveillance across protocols, networks, and services—provides effective support for centralized telecom maintenance and management, network optimization, service quality monitoring, business model analysis, network planning, and rapid fault localization and troubleshooting. Additionally, Fujian Mobile needs to perform various statistical analyses on signaling data. However, due to the massive data volume, traditional databases cannot meet the real-time computing requirements.

As China Mobile Fujian's subscriber base grows and services become increasingly complex, the core network's signaling data volume is expanding rapidly. To build a signaling monitoring system that enables fault localization and troubleshooting along with complex analysis of massive data—such as network hotspot analysis—a distributed-architecture MPP database is urgently needed. This system will allow operators to quickly identify and locate core network issues, thereby improving user satisfaction with network quality.

The data sharing layer of Fujian Mobile's Signalling Monitoring Platform is built on two GBase 8a MPP Cluster clusters, each with 6 nodes, totaling 12 nodes. One cluster primarily handles fast query location and statistical data processing, while the other cluster mainly stores various signalling data aggregated by the front-end acquisition gateway and statistical data from other systems. GBase 8a MPP Cluster, based on a modern Shared-Nothing MPP cluster architecture, performs comprehensive, in-depth analysis of network signalling, providing statistical value and report data ingestion, querying, and management functions. The GBase 8a cluster features an online dynamic horizontal scaling mechanism to meet the ever-growing demands of signalling data processing. The Fujian Mobile Signalling Monitoring System consists of 14 servers: 2 loading machines plus 12 nodes. Each cluster node serves as both a compute node and a management node, eliminating the need for a dedicated Master node. The current data volume is 36 TB.

Unified Network-Wide Signaling Monitoring and Processing: Supports real-time correlation processing for major events, delivers end-to-end fault localization and service analysis across the entire network, and enables rapid detection and pinpointing of faults and hotspots to improve user satisfaction with network quality.

Low Cost, High Efficiency: Built on cost-effective x86 PC servers for a superior price-performance ratio.

High Availability: A security group-based backup strategy ensures uninterrupted service continuity, even if a node fails.