Prometheus is free software which facilitates metric gathering, querying and alerting for a wealth of different third-party languages and applications. It also provides integration with Kubernetes for automatic discovery of supported applications.
Visualization and querying can be done through its built-in expression browser or, more commonly, via Grafana.
An environment being administered or analyzed by Prometheus can include current and past metrics exposed by kdb+.
The following are potential use cases for the interface, this is by no means an exhaustive list
- effects from version upgrades (e.g. performance before/after changes)
- alerts when your that a licence may be due to expire
- bad use of symbol types within an instance
- provides a script for useful general metrics that can be extended if required
- allows correlations between different instances, metrics, exporters and installs to be easily identified
Some caveats regarding where this interface in its current iteration can be used
- This interface does not provide service discovery. Prometheus itself has support for multiple mechanisms such as DNS, Kubernetes, EC2, file based config, etc., to discover all the kdb+ instances within your environment.
- You may need to extend this script to provide more relevant metrics for your environment. Please consider contributing if your change may be generic enough to have a wider user benefit.
- General machine/Kubernetes/cloud metrics on which kdb+ is running. Metrics can be gathered by such exporters as the node exporter. Metrics from multiple exporters can be correlated to provide a bigger picture of your environment conditions.
In Prometheus metrics refer to the statistics being monitored. Within Prometheus are different forms of metric. The exposure of these metrics from a kdb+ session allows for the monitoring a kdb+ process with Prometheus.
There are four types of metric in Prometheus:
counter gauge histogram summary
These are classified as either Single-value or Sample metrics
- Single-value metrics
gaugeare single-value metrics, providing a number per instance.
When updating a single-value metric, a single number will be modified. On a request, this number will be reported directly as the metric value.
- Sample metrics
summaryare aggregate metrics, providing summary statistics (defined by the metric params) per instance.
When updating a sample metric, a list of numeric values will be appended to. On request, this list will be used to construct the metric values, depending on the metric type and params.
The interface is currently available as a beta version under an Apache 2.0 licence and is supported on a best-efforts basis by the Fusion team. The interface is currently in active development, with additional functionality to be released on an ongoing basis.