Please enable JavaScript.
Coggle requires JavaScript to display documents.
Prometheus is an open-source monitoring and alerting toolkit. It collects…
Prometheus is an open-source monitoring and alerting toolkit. It collects and stores metrics as time-series data alongside an optional key/value pairs called labels (a.k.a tags).
Metric types
Counter
It is a cumulative metric that represents single a monotocity counter whose value can only increase and reset to zero when it restarts, like number of served requests.
Gauge
It is a metric that represents a single numeric value that can go up and down like memory usage. It can also represent counters that can go up and down such as number of parallel requests.
Histogram
It is a distributed measurements of data occurrence. Metrics of histogram type are stored as buckets of occurrences when data is scraped.
-
-
Features
Multi-dimensional data model: time series are identified with metric names and pairs of key/value labels that can give several perspectives (dimensions) of a given time serie.
-
No reliance on distributed storage. Prometheus stores data in the file system of its own instance. While it makes easy to configure Prometheus server, it becomes difficult to scale out instances horozontally.
-
Time series are collected via pull model over HTTP. Instances should provide HTTP endpoint with metric data which are scraped from time-to-time interval by Prometheus.
Pushing time series is allowed via Pushgateway, which allows short-lived services to push time series to a Pushgateway which Prometheus can scrape.