An LLM-based system for automating chaos engineering скачать в хорошем качестве

An LLM-based system for automating chaos engineering

As communication networks grow in scale and complexity, even partial failures can impact the entire network. However, predicting failures in advance is difficult. We are researching network fault tolerance. As part of this, we developed a technology using Chaos Engineering, which intentionally injects failures, observes system behavior, and improves resilience. Chaos Engineering for a system follows four steps: 1. Hypothesis: Define steady state of the system & failure scenarios. 2. Experiment: Inject simulated failures. 3. Analysis: Observe and analyze system response. 4. Improvement: Reconfigure the system. While experiment and analysis automation has advanced, hypothesis and improvement still require expertise in networks and chaos engineering. To solve this, we developed a system that fully automates Chaos Engineering with Large Language Models (LLMs). This video introduces NTT’s LLM-based Chaos Engineering automation technology.