What are the two types of metric collection (push and pull)?

There are two main approaches in monitoring and metric collection: push and pull.

• Push: the application or service independently sends metrics to the monitoring server (for example, via HTTP POST). This is convenient when services are not directly accessible for polling or when data needs to be sent on a schedule or upon events.

• Pull: the monitoring server periodically polls the application, requesting metrics (for example, via HTTP GET). This approach is often used in systems like Prometheus, where exporters open an endpoint with metrics, and monitoring reads them.

The choice depends on the architecture and monitoring requirements. Pull ensures data freshness and ease of service discovery, while push offers flexibility and the ability to work under limited network conditions.