From 2931db9f3d0fa85c98490c17bb89054f09b5d2e4 Mon Sep 17 00:00:00 2001 From: lavado Date: Tue, 12 Nov 2019 11:43:07 +0100 Subject: [PATCH] VNF Indicator collection using Prometheus exporters Change-Id: I7cdf1fde2c0dbe4791d3c9250e10d6775d42d626 --- .../vnf_indicator_collection_exporters.md | 35 +++++++++++++++++++ 1 file changed, 35 insertions(+) create mode 100644 Release8/vnf_indicator_collection_exporters.md diff --git a/Release8/vnf_indicator_collection_exporters.md b/Release8/vnf_indicator_collection_exporters.md new file mode 100644 index 0000000..db98a47 --- /dev/null +++ b/Release8/vnf_indicator_collection_exporters.md @@ -0,0 +1,35 @@ +# VNF Indicator collection using Prometheus exporters + +## Proposers + +- Gianpietro Lavado (Whitestack) +- Gerardo Garcia (Telefonica) +- Francisco Javier Ramon (Telefonica) + +## Type + +Feature + +## Target MDG/TF + +SA + +## Description + +VNF indicators are gathered today using Juju Metrics, which MON collects through N2VC. +Juju metrics present some limitations: +- Collection interval is fixed to 5m +- Metrics can only be collected through CLI + +The original proposal to replace this with regular, periodic primitives, has been discarded as +primitives are 'blocking', so a metric collection primitive could delay a configuration primitive. + +A mechanism is needed to collect indicators through SNMP, REST API and other methods, ideally: +- Adding 'exporters' to store metrics directly to the existing Prometheus TSDB. +- Trying to mantain alignment with current ETSI standards (SOL002, SOL006) + +## Demo or definition of done + +VNF Indicator collection through SNMP (ant other methods) can be specified at the VNFD, with a +variable collection period, which will make the system enable the corresponding exporter and +show the metric at a Network Service dashboard in Grafana. \ No newline at end of file -- 2.25.1