[Power BI] Anomaly management report

📝 Scenario

As a FinOps practitioner, I need to detect and respond to anomalies in order to minimize unexpected charges

💎 Solution

Required for the release:

• Initial Storage report
• Fix the DateTable.Date relationship to handle when reporting grain is set to months.
• Change the lookback to be 28 days (rather than 1 month, which may be a different number of days)
• Make the lookback period configurable (e.g., slicer or parameter)
• Make the min delta amount configurable (e.g., slicer or parameter)
• Make the max delta optional and configurable (e.g., slicer or parameter)
• Create KQL version of the report
• Update marketing page
• Document report pages
• Update the changelog

Stretch goals:

• Add KQL anomaly detection function
• # anomalies per month KPI
  • Formula: Total number of anomaly alerts "ON" per month (count of alerts generated by an anomaly detection system)
  • Assumption: each anomalies detect an unexpected or unforecasted cloud cost events in a timely manner deviating from the initial budget. Anomaly detection identified cost increases in an attempt to avoid surprise charges when the monthly bill arrives.
  • Objective: To measure the number of unpredictable variation events in the actual consumption that are detected by the system or the FinOps team
  • Value: This metric provides valuable insights into the effectiveness of the anomaly detection system and the organization's ability to proactively identify and address potential issues. It serves as an important tool for maintaining operational awareness and making informed decisions about resource allocation and system improvements.
    • (Consistently) High alert count: indicate that the anomaly detection is vigilant and identifying potential issues regularly. It is crucial to review each alert's significance and prioritize investigations based on potential impact.
    • (Consistently) Low alert count: suggests that the systems are operating within expected norms, which can be a positive sign. However, it's important to ensure that the system is not overlooking anomalies (review the anomaly detection criteria) and improve sensitivity.
• Dismissed anomalies KPI
  • Formula: Total anomalies dismissed over the total anomalies alerted per month
  • Objective: To measure the percentage of Anomalies for which, having been flagged by the System, it has been decided that corrective measures should not be taken.
  • Value: The metric provides valuable insights into how effectively the organization is responding to anomaly alerts. It serves as an indicator of operational efficiency and risk mitigation efforts, helping to prioritize improvements in the anomaly detection and resolution processes. With this KPI we obtain information about the efficiency of the deviation alert system.
    • Low Ratio: suggests that you are effectively resolving most anomaly alerts. The company should continue to focus on proactive anomaly management, but also monitor trends to ensure that resolution rates remain high.
    • High Ratio: indicates that a substantial number of anomaly alerts are not being addressed. The company should investigate the reasons behind this, which may include resource constraints, inefficient processes, or a high volume of false positives.

🙋‍♀️ Ask for the community

We could use your help:

1. Please vote this issue up (👍) to prioritize it.
2. Leave comments to help us solidify the vision.

Start by reviewing the existing Initial Storage report and the unchecked items in the Solution checklist. Completion means implementing the required report, lookback, configuration, KQL, marketing, documentation, and changelog updates, with stretch goals handled separately.