I See Metrics: Anomaly Detection on OpenShift

Workshop Abstract

Monitoring with Prometheus is becoming a common pattern for cloud native applications. But knowing what metrics to record and how to find relevance is a challenging task. How can AI-based solutions help us automate the process of identifying and alerting when metrics behave differently and also augment human introspection?

In this workshop you will learn how to use the Prometheus Anomaly Detection framework to enhance your metrics automation. We will walk you through the process of deploying and operating your own anomaly detection framework. You will walk away from this workshop knowing how to:

• Setup a sample application to generate metrics
• Configure Prometheus to collect the metrics
• Use Python (Prometheus API client library) to transform metrics into a suitable format
• Train machine learning models to perform time series forecasting
• Use Grafana to create insightful dashboards and setup alerts

Workshop Content

The repository contains the following contents:

• a doc folder, containing:

  1. abstract.md the workshop abstract.

• a slides folder containing:

  1. rendered rendered pdf file of the slides
  2. source containing the source for the slides, as a powerpoint

• a source folder, containing:

  1. Jupyter notebooks required to give the workshop
  2. data needed for the workshop
  3. a README.md file, explaining all of the contents of the source folder

• a workshop.yaml file, holding the following machine readable metadata:

  • name
  • authors
  • workshop duration
  • source repo
  • keywords

Workshop Duration: 90 minutes

Source: https://github.com/AICoE/prometheus-anomaly-detector

Keywords: Prometheus - OpenShift - Machine Learning - AI - Jupyter - Python - Grafana - Metrics - Time series