This is an ape small talk 54 original share

Take a look at Prometheus. According to the official documentation about https://prometheus.io/docs/introduction/overview/ to learn about the Prometheus is an open source system monitoring and alarm kits, by time-series data based on Http pull way, Supports a variety of charts and interfaces.

As shown in the official framework diagram, Prometheus consists of exporters, Pushgateway, Prometheus Server, AlertManager, and Prometheus Web UI.

I have a feeling that if we keep talking like this, you’ll probably walk away in frustration, so if you want to go into details about what each kit does, you might as well post an image based on the official architecture.

As shown in the figure above, I simply divided the ecosystem of Prometheus into three layers: data source, data collection and processing, and data application.

Data source layer, which produces data in two main forms.

  • One way is to integrate Prometheus by adding the application to the Clinet Lib library and then pushing metrics data to PushGateway. For example, when Prometheus monitors Flink, it needs to add the flink-metrics-Prometheus_XXXX. jar library to support it.

  • The other is to run a series of XX_exporters provided by Prometheus to collect monitoring indicator data. For example, Node_exports can be used to collect CPU, memory, disk, AND I/O information of a server. (This is my friend from PGOne.)

Data collection processing layer: Prometheus server pulls and stores monitored indicator data from the data source layer through Http pull method; Then run defined alarm rules to push alerts to the Alertmanager at the data application layer. Thinking about Prometheus Server, it is necessary to maintain a monitoring target, otherwise how to know where to pull data? Practice and then experience it, first know what this layer is doing.

Data application layer, I roughly divided this area into alarm management and monitoring data display two applications. Alarm management refers to that the Alertmanager processes the received alerts and generates alarms based on the configuration. Monitoring data presentation is mainly used for monitoring indicator data presentation, the commonly used component is Grafana (what is this? Grafana will catch up later).

To this should monitor a brother Prometheus has a new understanding, later in the selection of technology might as well consider him a little, this is mainly to understand the concept, about the actual combat part of let’s start to speak alone.

Well, if this brief sharing is still pleasing to the eye, might as well carry forward the style, move to show a “watching”, so that more people progress together.