As you can see, a list of "NFVI metrics" is defined first at the VDU level, which contains an ID and the corresponding normalized metric name (in this case, `cpu_utilization` and `average_memory_utilization`). Normalized metric names are: `cpu_utilization`, `average_memory_utilization`, `disk_read_ops`, `disk_write_ops`, `disk_read_bytes`, `disk_write_bytes`, `packets_received`, `packets_sent`, `packets_out_dropped`, `packets_in_dropped`
Not all metrics can be collected from all types of VIMs, the following table shows which metrics are supported by each type of VIM:
| Metric | Openstack | Azure | GCP |
| ------ |:---------:|:-----:|:-----:|
| cpu_utilization | X | X | X |
| average_memory_utilization | X || X |
| disk_read_ops | X | X | X |
| disk_write_ops | X | X | X |
| disk_read_bytes | X | X | X |
| disk_write_bytes | X | X | X |
| packets_in_dropped | X |||
| packets_out_dropped | X |||
| packets_received | X || X |
| packets_sent | X || X |
Available attributes and values can be directly explored at the [OSM Information Model](11-osm-im.md). A complete VNFD example can be downloaded from [here](https://osm.etsi.org/gitlab/vnf-onboarding/osm-packages/-/blob/master/hackfest_basic_metrics_vnf).
@@ -108,6 +108,21 @@ The following figure illustrates graphically the specific workflows for metric a
Regarding to resource consumption metrics, three types of VIM are currently supported: Openstack (Gnocchi and Ceilometer back-ends), Azure and GCP. The metrics that OSM can collect from the VNFs deployed in those VIMs are shown in the table below:
| Metric | Openstack | Azure | GCP |
| ------ |:---------:|:-----:|:-----:|
| osm_cpu_utilization | X | X | X |
| osm_average_memory_utilization | X || X |
| osm_disk_read_ops | X | X | X |
| osm_disk_write_ops | X | X | X |
| osm_disk_read_bytes | X | X | X |
| osm_disk_write_bytes | X | X | X |
| osm_packets_in_dropped | X |||
| osm_packets_out_dropped | X |||
| osm_packets_received | X || X |
| osm_packets_sent | X || X |
The metrics, as acquired by OSM, might not be useful per-se, since they are not related to OSM objects. For instance, the VM status or the VM resource consumption metrics are only useful when they relate to a VNF instance or a NS instance. [Prometheus recording rules](https://prometheus.io/docs/prometheus/latest/configuration/recording_rules/) allow to derive metrics from existing metrics, computing them and saving the result as a new set of time series. The following figure shows a screenshot of the metrics derived in Prometheus.
