Neusoft Group uses Lindorm to power smart operations and maintenance (O&M) across its digital systems for public services, telecommunications, automotive, and healthcare customers. With Lindorm as the unified storage engine, Neusoft consolidates metrics, logs, traces, and network packets from a large number of distributed endpoints into a single platform—cutting storage costs by 40% and processing 120 million monitoring data points every day.
Figure 1. Lindorm-based IT O&M and monitoring system
About Neusoft
Founded in 1991, Neusoft is the first publicly listed software company in China. With nearly 20,000 employees across R&D centers, sales networks, and service offices in more than 60 cities, Neusoft also operates subsidiaries in the United States, Japan, and Europe.
Neusoft serves enterprises in smart cities, healthcare, intelligent vehicles, and software engineering. PricewaterhouseCoopers has ranked Neusoft among the top 100 global software enterprises for four consecutive times. The company is also recognized as one of the top 20 most globally competitive Chinese companies, one of the top 50 global challengers in China, the most appreciated knowledge-based enterprise in Asia, and the best employer in the Asia-Pacific region.
The challenge: monitoring data that outgrew single-model storage
Modern digital services generate several distinct types of monitoring data simultaneously: time series metrics from endpoints, log streams from applications, distributed traces from microservice calls, and network packet captures from infrastructure links. Storing and analyzing these data types together is fundamentally difficult because each has different write patterns, query shapes, and retention needs.
Neusoft faced two compounding problems as its customer base grew:
Write-side pressure. Mobile terminals, IoT devices, and network sniffers across China generated concurrent write requests that single-model engines could not absorb. The existing system struggled with write concurrency, dropping data under peak load.
Analysis-side fragmentation. Because Neusoft relied on separate engines for different data types—including Round Robin Database (RRD), OpenTSDB, and Elasticsearch—cross-model analysis required manually correlating data across systems. A fault event involving both a metric spike and a log anomaly meant querying two or more separate systems and stitching the results together. This made root-cause analysis slow and error-prone.
The economics were also unsustainable. Low-value historical monitoring data still needed to be retained for compliance and trend analysis, but storing large volumes of it in high-performance engines was not cost-effective. Self-managed open-source clusters added further strain: unstable performance and high ongoing maintenance costs consumed engineering time that should have gone to product development.
The scale of these problems is well documented across the industry. According to Forrest research, 57% of enterprises experience at least one critical application error per week, and 28% have errors that occur every day. Forrest also reports that 42% of enterprises say application performance problems directly reduce revenue, and 52% of O&M teams scatter their investment across multiple monitoring and management tools as a result.
Solution: Lindorm as the unified O&M data platform
Neusoft evaluated the requirements of customers across telecommunications, automotive, and public services sectors and chose Lindorm—Alibaba Cloud's cloud-native multi-model database—as the core storage engine for its O&M platform.
Figure 2. A Lindorm-based solution for multi-model O&M data storage and analysis
Use cases
Real-time monitoring dashboards. Operations teams query live metrics and log data through dashboards, tracking application health across all customer deployments without switching between systems.
High-volume log retrieval. When an incident occurs, engineers search across billions of log records to narrow down the affected service, time window, and error pattern—all in a single query.
User behavior tracking. Product and operations teams trace individual user sessions across mobile and cloud touchpoints to identify performance degradation affecting specific segments.
Fault tracing and root-cause analysis. Correlated analysis across metrics, logs, and traces in one platform reduces mean time to resolution. Engineers identify the root cause without manually joining data from separate systems.
AI-powered anomaly detection. Lindorm's storage and retrieval capabilities feed real-time monitoring data into anomaly detection models, enabling proactive alerting before users are affected.
Results
Neusoft has deployed this architecture on Alibaba Cloud for a vehicle company, collecting user digital footprints and experience data from customer terminals worldwide. The system handles access requests from tens of thousands of daily active users and ingests 120 million tuples of real-time monitoring metric data every day.
Storage and system maintenance costs are 40% lower than what Neusoft paid with the previous engine stack.
Figure 4. O&M big data visualization dashboard 1
Figure 5. O&M big data visualization dashboard 2
Figure 6. O&M big data visualization dashboard 3
Figure 7. O&M big data visualization dashboard 4
Why Lindorm
| Capability | Impact |
|---|---|
| Multi-model storage (metrics, logs, traces, network packets) | Eliminates cross-system joins; enables integrated analysis in a single query |
| High-throughput concurrent writes | Absorbs peak write load from thousands of simultaneous endpoints without data loss |
| 99.99% data availability | Reduces risk of monitoring blind spots caused by data loss |
| Low-cost cloud storage for high-volume, low-value data | Retains all historical monitoring data without inflating storage costs |
| Fully managed service | Frees engineering teams from cluster maintenance; eliminates unstable open-source cluster issues |
| Connectivity from any endpoint | Simplifies network configuration across mobile, cloud, and data center environments |