It can be seen that there are many products that store and analyze indicators separately, among which Pro is the most widely used. The general storage model of the collected indicators is a typical time series data model. In the cloud native scenario, there are many containers, many monitoring objects, and large writing concurrency. In addition, a single indicator has obvious time attributes. Some analysis services may require long-term data storage, so the data volume is also relatively large. For example, a business on Huawei Cloud may generate 20TB of indicator data and 80TB of log data every day. These data often have a timeliness, and the data can often expire and be eliminated periodically. In the production environment, data is continuously generated.
For example, we need to use the system's resource indicators to japan mobile phone number list monitor alerts and detect anomalies, to help operation and maintenance personnel find problems early, and to provide direction for rapid problem analysis and root cause location. These business scenarios have some common characteristics, mainly the following:
• It does not update or delete data
• Pay more attention to recent data and less attention to historical data
• Mainly based on aggregate queries within time range
• High query response speed is required, and sensitive monitoring capabilities are required
The Prometheus technology ecosystem is active. What are its advantages and disadvantages ?
Currently, Prometheus is the most widely used monitoring platform in enterprises. It has the following advantages:
• Complete indicator data model: Each time series consists of a metric name and a set of labels
• Powerful query capabilities: Flexible query language PromQL helps users query indicator data
• Easy to integrate: developed based on golang , compatible with multiple platforms, fast and user-friendly
• Cloud-native friendly: Deep support for Kubernetes, flexible service discovery mechanism helps effectively monitor large-scale kubernets clusters
• Various client libraries: Client libraries in multiple programming languages make it easier for developers to embed monitoring code in their applications
• Visualization and integration: It can be integrated with data visualization tools such as Grafana to provide rich charts and dashboards to help users intuitively understand monitoring data
• High availability solutions: Community high availability solutions such as Cortex and Thanos help to monitor data on a large scale.