AIOps in Action: Incident Prediction and Root Cause Automation Training Course

Introduction to Predictive AIOps

• Overview of predictive analytics within IT operations.
• Identifying data sources for prediction, including logs, metrics, and events.
• Key concepts in time-series forecasting and anomaly pattern recognition.

Designing Incident Prediction Models

• Labelling historical incidents and system behaviour for training.
• Selecting and training models, such as LSTM, Random Forest, or AutoML.
• Assessing model performance and managing false positives.

Data Collection and Feature Engineering

• Ingesting and aligning log and metric data for optimal model input.
• Extracting meaningful features from both structured and unstructured data.
• Managing noise and missing data within operational pipelines.

Automating Root Cause Analysis (RCA)

• Utilising graph-based correlation to map services and infrastructure.
• Applying ML to infer probable root causes from event chains.
• Visualising RCA outcomes through topology-aware dashboards.

Remediation and Workflow Automation

• Integrating with automation platforms, such as Ansible or Rundeck.
• Triggering automated actions like rollbacks, restarts, or traffic redirection.
• Auditing and documenting automated interventions for compliance and insight.

Scaling Intelligent AIOps Pipelines

• Applying MLOps to observability: focusing on retraining and model versioning.
• Executing real-time predictions across distributed nodes.
• Best practices for deploying AIOps in production environments.

Case Studies and Practical Applications

• Analysing real-world incident data using predictive AIOps models.
• Deploying RCA pipelines using both synthetic and production data.
• Reviewing industry use cases, including cloud outages, microservices instability, and network degradations.

Summary and Next Steps