01 Current State and Pain Points of Hardware Monitoring

Enterprises typically need to monitor and manage the operational status of various resource instances from a business perspective to ensure efficient operation. These resource instances include, but are not limited to, servers, network devices, security appliances, and storage devices. Understanding the status of these resources is not only the foundation for maintaining IT infrastructure but also a critical element in ensuring business continuity and performance stability.

Real-time monitoring of hardware devices and their metric performance helps identify potential issues promptly and take corrective actions, thereby minimizing downtime. For example, by monitoring key metrics such as CPU load, memory utilization, and hard disk I/O performance, enterprises can anticipate and address resource bottlenecks and optimize performance configurations. Additionally, monitoring bandwidth usage and latency of network devices ensures smooth data transmission and prevents business disruptions.

However, when implementing hardware monitoring, enterprises often face the following pain points:

• Wide variety of hardware devices

Hardware devices are not only categorized by type—network devices, physical servers, storage, security appliances, and more—but are further classified within each category by different vendors and models. Since an enterprise's hardware environment is rarely built all at once, organizations often find themselves needing to monitor hundreds of different hardware device models. How to adapt to and monitor all device models has become the greatest challenge in hardware monitoring.

• Lack of data processing capabilities

For hardware devices, the monitoring data collected through various out-of-band protocols is often relatively fixed, and many core metrics cannot be obtained directly from the device. For instance, bandwidth utilization—if collected via the SNMP protocol—the device only exposes total port traffic metrics rather than rate metrics directly. In this case, a Rate function calculation must be applied to the collected data to derive the port rate that users actually care about.

• Absence of topology capabilities

In hardware monitoring scenarios, the focus is rarely on a single physical device alone. Instead, it is more often on the overall network architecture and the link status between devices. Beyond comprehensive detection and alerting capabilities, visual topology capabilities are needed to build and display such information based on topology views.

• Lack of a unified monitoring tool

As a standalone monitoring domain, there are already many products in the IT O&M field that specialize in hardware monitoring. However, for enterprises, using a mix of multiple monitoring products significantly increases management costs. A feature-rich, centralized monitoring product can undoubtedly greatly improve monitoring and IT O&M efficiency.

CanWay BlueWhale Full-Stack Observability Center (hereinafter referred to as the "Monitoring Center") is a powerful enterprise-grade IT O&M monitoring tool designed to meet the complex needs of various enterprises. It offers out-of-the-box hardware monitoring capabilities for real-time monitoring of hardware device status and performance. This article will introduce the Monitoring Center's hardware monitoring solution, covering the entire process from data integration, data detection, to data visualization.

02 Monitoring Center: A Comprehensive Hardware Monitoring Solution

Data Integration

For hardware devices such as servers, network devices, security appliances, and storage devices, the Monitoring Center supports multiple out-of-band protocol integrations and various data collection types: metric collection and log collection based on SNMP and IPMI protocols. These diversified collection methods meet the specific needs of different enterprises, ensuring the flexibility and adaptability of the monitoring system. Meanwhile, the Monitoring Center can efficiently monitor device health, including critical metrics such as hardware status, CPU load, memory usage, and port traffic, supporting data collection intervals as low as 10 seconds to meet enterprise requirements for second-level hardware resource monitoring.

IPMI Protocol-Based Collection

The system has a built-in IPMI protocol plugin that can interface with any hardware server with IPMI enabled—such as rack servers and blade servers—to retrieve relevant metrics without any modification or adaptation.

SNMP Protocol-Based Collection

The Monitoring Center provides a no-code SNMP plugin creation capability. As long as you have the corresponding device's MIB library or specific metric OIDs, you can quickly create collection plugins for the target device through page-based configuration, customizing the retrieval of device status information and performance metrics. These devices primarily include servers, network devices, storage devices, and load balancing devices.

It also comes with a batch of out-of-the-box, standardized plugins covering mainstream device models on the market. In most cases, these plugins can fulfill enterprise monitoring needs without additional plugin development.

Additionally, the Monitoring Center provides detailed metric documentation and best-practice configuration guides to help users thoroughly understand the meaning of each metric and how to configure monitoring policies. These documents detail each metric's definition, normal ranges, and other information to ensure users can accurately interpret monitoring results.

Log-Based Collection

Furthermore, the Monitoring Center supports data collection through device-generated logs. It can collect log files from multiple sources via the Syslog protocol; supports configurable filtering rules so that collection only occurs when filter conditions are met, reducing bandwidth consumption during transmission; and provides rich data cleansing capabilities. This flexible log management mechanism not only meets business scenario requirements and improves data collection efficiency but also ensures the comprehensiveness and accuracy of information.

Data Detection

The Monitoring Center supports multiple metric detection algorithms to flexibly address different monitoring needs. The system also features powerful metric calculation capabilities, allowing users to define and detect custom derived metrics for a more precise reflection of system performance and health status. Additionally, for hardware log monitoring, the Monitoring Center provides data detection through log keywords. This multi-layered, multi-dimensional monitoring approach helps users achieve more efficient IT O&M management.

• Metric Detection: Supports eight anomaly detection algorithms—static thresholds, year-over-year (advanced), period-over-period (advanced), year-over-year (simple), period-over-period (simple), year-over-year amplitude, period-over-period amplitude, and year-over-year range—as well as no-data alerts, enabling single-metric or multi-metric detection capabilities with recovery condition settings.

• Metric Calculation: Supports metric calculation capabilities, including metric function computation and multi-metric expression operations to handle a wide variety of monitoring scenarios.

• Derived Metrics: Supports predefined calculation rules to generate new metrics from existing ones. Derived metrics can be directly consumed in monitoring policies, dashboards, and other components. For metrics that consistently require calculation, users can create derived metrics to reduce subsequent configuration costs.

• Log Keyword Detection: The keyword detection function allows users to retrieve qualifying log records through custom search statements. Combined with the metric detection algorithms described above, it enables comprehensive monitoring of log keywords.

Data Visualization

The Monitoring Center provides multiple visualization methods to meet hardware monitoring requirements from different IT O&M perspectives. Business managers can view the overall operational status of the entire IT asset network topology from a business perspective to grasp the macro-level system health, with the ability to drill down directly from the network topology to specific resource instances. Technical IT O&M personnel can directly access device metric views and alert management information by resource instance, or locate and resolve specific issues through hardware log search. Through various data visualization methods, personnel in different roles can monitor and manage hardware devices more efficiently.

1. Network Topology

The Monitoring Center provides a network topology feature that allows users to customize topology diagrams and associate them with actual instances. Users can intuitively display the status and alert information of each device within the topology, staying informed of network conditions in real time and improving the efficiency of troubleshooting and management.

2. Resource Instance Details

The Monitoring Center offers multi-perspective hardware monitoring views to meet the needs of different IT O&M roles. Users can view metric views of resource instances from a resource model perspective to understand the health status of each resource instance, facilitating problem identification and resolution.

3. Hardware Log Search

For hardware log visualization, the system supports log query and display through Elasticsearch native syntax and regular expressions, providing near-real-time search capabilities. Users can perform full-text search, cross-business search, and data-masking search to meet requirements across different scenarios. Additionally, the system offers a one-click conversion to monitoring policy feature, simplifying the monitoring configuration process. Combined with real-time log and context capabilities, users can perform troubleshooting and problem analysis more efficiently.

4. Dashboards

Furthermore, the Monitoring Center supports metric or log configuration through common dashboard components, enabling flexible customization and display of various key information. Through dashboard configuration, users can not only intuitively monitor system operational status but also analyze the performance of different resource instances in real time. This configuration approach gives dashboards a high degree of scalability and adaptability, meeting various business needs and providing robust support for system optimization and maintenance.

03 Conclusion

After thoroughly analyzing the current state of hardware monitoring, its pain points, and the comprehensive solution provided by the Monitoring Center, we can clearly see that the importance of hardware monitoring in enterprise management is increasingly prominent. With its powerful capabilities and flexible adaptability, the Monitoring Center provides enterprises with a complete infrastructure monitoring solution. By adopting the Monitoring Center, enterprises can more efficiently manage and maintain their IT infrastructure, ensuring business continuity and performance stability, and laying a solid foundation for long-term intelligent IT O&M and business growth.