Zabbix meets AI–The power of opensource monitoring
summary
Shows a concrete integration: a trained autoencoder LSTM ingests multivariate time‑series pulled via the Zabbix API, learns normal cross‑metric and temporal patterns, and emits a live anomaly probability as a new Zabbix item. Training uses historical windows with enrichment (time features) and normalization, testing shows the model flags a web‑shop compromise and an earlier probe the humans missed, and reconstruction error metrics (reconstruction error (MAE, MAPE)) drive a 50%‑style trigger; runner scales horizontally for real‑time inference.
AI-written summary · about
chapters (9)
- 0:00Introduction to AI integration
- 2:04Anomaly detection concepts
- 6:13Machine learning fundamentals
- 10:32Data types and network architecture
- 14:26Autoencoder and LSTM cells
- 17:51Data preparation for AI
- 21:33Building blocks for integration
- 23:30Practical implementation results
- 28:10Summary and conclusion
related talks
25mMonitoring AI with AI
12mUnleashing the Power of AI for Performance Monitoring& Anomaly Detection
19mThe Rise of AI in Anomaly Detection
31mBuilding Trustworthy AI: Open Source Monitoring in Action
28mHow Machine Learning Can Achieve Data Quality Monitoring At Scale
29mAI Driven Observability based on Open Source
39mMastering real-time anomaly detection with open source tools
58mZabbix: Recipes for High-Performance Monitoring