Customer Overview
A leading rural commercial bank in Jiangsu Province operates at leading scale in the province, with branches covering urban and rural areas. Long committed to supporting agriculture and small businesses, agriculture-related and small and micro enterprise loans make up over three-quarters of its total loans, with small loans of RMB 10 million (inclusive) or less exceeding RMB 100 billion. Its long-term credit rating remains AAA. Years of development have built a service network spanning urban and rural areas across the province — an important force in regional financial services for agriculture, rural areas, farmers, and small and micro enterprises.
On the IT side, the bank manages 20+ core business systems, covering key links including counter services, credit, payment clearing, core accounting, ESB service bus, and UnionPay / CNAPS / Super Online Banking. As a regional legal-person bank, its digital transformation has moved from system construction into a new stage of data-driven, intelligent IT operations, and refined governance. That's why the bank's Black-Box Observability and Intelligent Alert Analysis practice, co-built with CanWay, was selected for the 5th "Dingxin Cup" Digital Transformation Application Collection (Intelligent IT Operations category) — a benchmark for observability in the financial industry.

Industry Background: Business Continuity Regulation Drives IT Operations Upgrades
Banking business continuity management is anchored in the Guidelines on Business Continuity Supervision for Commercial Banks— requiring rapid recovery of critical businesses after disruptions, reasonable RTO/RPO targets, and multi-layered contingency plans. For this rural commercial bank, 7×24 stable operation of core links like payment clearing and credit directly affects regional financial order and social well-being. As monitoring scope and alert volume keep expanding, adding more monitoring rules alone can no longer sustain localization efficiency — a unified observability foundation and intelligent IT operations capabilities have become an essential gap to close for compliance and business continuity.
Business Challenges
As the bank's business scale keeps expanding and its IT architecture grows more complex, the existing IT operations system faces new management challenges on its journey toward proactive awareness and intelligent analysis:
01 Monitoring Systems in Silos, Alerts Without a Unified Foundation
Multiple monitoring systems — CanWay IT monitoring, Zabbix, BPC — run side by side. Alert formats, field structures, and identification methods all differ, forcing on-call staff to switch between systems repeatedly. High alert noise and frequent false positives leave teams stretched thin during alert storms, and fault localization relies heavily on individual experience.
02 Configuration Management Lacks an Authoritative Data Source, Decisions Lack a Basis
With over 20 systems, configuration item naming is inconsistent, topology relationships are missing, and business systems aren't mapped to infrastructure. Fault impact scope is hard to assess quickly, and change and production release risk control lacks data support.
03 Observability Stays at the Metric Display Layer, Key Business Links Lack Visibility
Critical payment links — ESB service bus, UnionPay, CNAPS, Super Online Banking — lack end-to-end black-box monitoring and visual dashboards. When faults occur, it's impossible to quickly determine whether the issue lies in the business system, network link, or service endpoint — making cross-system troubleshooting inefficient.
04 Weak Intelligent Analysis, IT Operations Knowledge Hard to Accumulate
Alert handling relies heavily on senior staff. Troubleshooting experience stays scattered across individuals, hard to standardize and transfer. Newcomer ramp-up is slow, and fault retrospectives lack a structured evidence chain.
05 Under Domestic Technology Substitution, Infrastructure Faces Adaptation and Migration Pressure
Upgrades like domestic databases, application sandboxes, and resource isolation have yet to form a unified plan — while security and compliance requirements keep rising.
Why CanWay
Facing these challenges, the leading rural commercial bank in Jiangsu Province evaluated market options thoroughly — then chose a deep collaboration with CanWay to co-build a black-box observability and intelligent alert analysis system on the CanWay platform.
Key Selection Factors:
Platform Alignment and Continuity:The bank already runs a CanWay ecosystem. CanWay products align naturally with its existing CanWay observability and standard IT operations capabilities — smooth onboarding and layered expansion, without replacing existing monitoring.
Integrated Observability Product Matrix:From configuration management and monitoring and alerting to operations visualization and automated handling — one system connects the closed loop: monitoring, correlation, analysis, resolution.
Financial-Grade Implementation and Domestic Technology Substitution:CanWay brings mature financial industry implementation methodologies and domestic technology stack deployment experience — supporting secure, compliant, stable operation in financial-grade offline isolated environments.
Solution and Implementation
Built on the CanWay PaaS, the project integrates BPC business transactions, Zabbix resource metrics, logs, and multi-platform alerts — then uses CMDB to link business, application, and resource objects, so monitoring, analysis, and resolution flow within a single system. Alert analysis follows a deterministic evidence + AI approach: the platform first runs metric calculation, link tracing, and evidence extraction, then hands confirmed facts to AI to produce human-readable anomaly explanations, impact assessments, and troubleshooting recommendations. Final conclusions and remediation actions remain subject to IT operations staff confirmation.
Overall Architecture and Core Capabilities:CanWay deployed and implemented the following core capabilities for the bank:

Core Capability 1: Multi-Source Data Fusion and a Unified Alert Foundation
Heterogeneous alerts from CanWay IT Monitoring, Zabbix, BPC, and other sources are mapped into a unified field model. Original payloads are retained, with idempotent deduplication by source + original event ID, generating alert fingerprints for problem patterns with identical semantics. A unified alert model and PostgreSQL fact database bring capabilities including original alert retention, alert aggregation, CMDB enrichment, historical statistics, co-occurrence analysis, explainable noise reduction, and audit trails.
Core Capability 2: CMDB Unified Resource Model and Panoramic Topology
A hierarchical model — application → service → service instance → host — is built in the CanWay Configuration Management Center, maintaining deployment architecture and upstream/downstream dependencies. Resources auto-associate with alerts and metrics, providing an authoritative data foundation for fault impact assessment and change risk control.
Core Capability 3: Automated Post-Alert Verification + Deterministic Analysis and Generative AI Collaboration
Raw alerts don't go straight to LLMs. Based on configuration, the system automatically traces evidence — transaction codes, requesters, service providers, return codes, minute-level trends, ESB nodes/IPs/ports. Rules and programs handle fact verification, metric calculation, and weak evidence filtering; AI then turns confirmed facts into easy-to-understand anomaly explanations, impact assessments, and verification recommendations. The principle: display only with evidence; no invalid information on screen.
Core Capability 4: A Human-Machine Collaborative Closed Loop Centered on Incidents
Scattered alerts become incidents. AI analysis generates structured reports stored in the database; after on-call staff confirm, the system links to the CanWay IT Automation Operations for controlled execution. The entire process is captured in the incident timeline and audit records — a complete incident handling chain: multi-source alerts merge into one incident, the platform verifies evidence and analyzes, staff confirm and execute under control, recovery is validated and archived, and accumulated experience feeds back into rules and the knowledge base.
Results
Through continuous development, the bank achieved significant improvements across the full chain — fault detection, localization, resolution — and in intelligent IT operations capabilities.
Efficient Results:

Validated by Three Typical Scenarios: Three Capability Leaps
1. IC Card Protocol Payment Timeout Localization:
The system automatically identifies anomalous transaction codes and traces ESB endpoint IP/port along upstream and downstream links — cutting localization time from about 30 minutes to about 2 minutes, with a complete, traceable evidence chain.
2. UnionPay Front-End Outgoing Transaction Success Rate Anomaly:
Transaction codes serve as the primary anchor, with issuing and acquiring banks as secondary anchors. The system independently separates the current direction from the paired direction — resolving direction confusion and unclear primary object challenges, and standardizing on-call analysis.
3. Multi-System Correlated Fault Governance:
Dozens of fragmented alerts are automatically organized into a single incident, with a significant improvement in noise reduction. On-call staff shift from reviewing alerts one by one to reviewing structured incident reports.
Capability Results: Agility and Domestic Technology Substitution Value
1. During core system debugging, performance bottlenecks were quickly localized — optimizing transaction response time standards of 60–80ms and shortening new business launch cycles.
2. For domestic technology substitution tuning, performance evaluation and roadmap selection for distributed databases and domestic processors were completed quickly — cutting the trial-and-error costs and cycles of adaptation.
3. During major event assurance periods, system anomalies were detected at second-level awareness — ensuring smooth operation through peak scenarios.
Extended Business Value
IT operations is no longer just reactive firefighting — it's proactive awareness, intelligent analysis, and closed-loop resolution. The unified observability foundation helps new architectures like cloud-native and microservices land smoothly. Accumulated full-chain observability data also supports domestic database migration and domestic substitution — while meeting compliance requirements for business continuity and audit traceability.






























