Metrics and Observations v3.5.0 (TR292G)

This document defines an ontology model for defining and using metrics and managing observations, such as measurements concerning the metrics.

Intent requirements are often based on setting target values for measurable properties such as a KPI value or a state being reached. For example, conditions contributing to intent expectations or context depend on them. These are in general metrics and observations about them. It is possible in intent models to use URIs that directly identify a node in the knowledge graph, where the value to be used is allocated and kept up to date. But this would mean that intent owners need the exact URI the measurement implementation of the intent handler uses for a particular measurement. This requires that intent handlers expose many details, and it also is limited regarding contextual information about the metric and its observation. For example, the time of latest value update or the scope of the value presented. An example of measurement scope is a sunset of resources or users included in the measurement. For example, a measurement of latency shall be the average over a defined set of users, but excluding other users. Managing these use cases requires a foundational model regarding metrics and dedicated vocabulary in the intent ontology.