Causal Analysis
Container Observability helps users analyze the root cause of the anomalies that triggered an alert. The RCA subsystem receives alerts and then follows the following process:
Select and execute a plan:
Using the information in the incoming Alert. For example, the type of Alert or the type of Object.
Decide on the point in the relationship graph to begin a 360 analysis.
Perform Assessments:
The plan executes a series of info-gathering from sources like the current status, configuration, logs, metrics, etc
Perform Decision Logic:
Find possible causal possibilities and collect data
Walk the topology watching the quality of findings and cost
Finalize and attach the RCA to the Alert:
Now the RCA is available to view and forwards to notification systems
In Container Observability, root cause analysis is explained using different visualizations.
Ishikawa (Fishbone) diagram:
The default visualization is the Ishikawa diagram (also called the Fishbone diagram).
The Ishikawa diagram comes along with annotated charts and graphs to understand the likely cause(s) of the anomaly.
Specialized diagrams:
For certain alerts, the RCAs are better represented by other diagrams as in the case of an SLO breach alert.
Related topics: