Key Technologies

Through a series of key technologies, Guanyun has built a complete technology stack from unified data modeling and real-time computing to intelligent query and analysis, delivering a unified, efficient, flexible, and scalable observability solution.

Unified Observability Data Model

Heterogeneous data such as Traces, Metrics, and Logs are unified under a single data model, enabling seamless cross-data-source correlation analysis and drill-down capabilities.

High-Performance Real-time Computing Engine

Leveraging Flink's stream processing capabilities, the platform achieves second-level aggregation and alerting for massive volumes of metrics.

Self-developed Unified Query Language NBQL

NBQL abstracts the underlying physical storage from the application layer, providing a concise, powerful, semantically rich, and extensible query interface based on logical models. This significantly lowers the barrier for data analysis by non-technical personnel while greatly enhancing platform openness.

Observability Data Lakehouse Architecture

Raw observability signal data is ingested into the lake, compensating for the feature loss caused by pre-aggregation in metrics. The innovative "Jianwei" multidimensional exploration engine can perform real-time holographic analysis across hundreds of dimensions simultaneously, helping users discover and locate issues more efficiently.

Hybrid Storage Architecture

Combining ClickHouse with the self-developed NBFS (Nabaichuan File System), structured and unstructured data are stored separately. Detailed data such as traces is stored separately from structured metric data, better balancing storage costs (detailed data is characterized by large volume and low access frequency, making it more suitable for low-cost cold storage).