Causal Inference-Based Root Cause Analysis for Online Service Systems with Intervention Recognition

Understand

Fault diagnosis is critical in many domains, as faults may lead to safety threats or economic losses.

• In the field of online service systems, operators rely on enormous monitoring data to detect and mitigate failures.
• Quickly recognizing a small set of root cause indicators for the underlying fault can save much time for failure mitigation.
• In this paper, we formulate the root cause analysis problem as a new causal inference task named intervention recognition.