EventOrOutage
Identifies potential external events like holidays or major happenings that may be causing traffic drops, helping SREs distinguish them from actual outages.
EventOrOutage leverages Large Language Models (LLMs) to assist Site Reliability Engineers (SREs) in determining whether a decrease in traffic is due to external events, such as holidays, elections, or sporting events, rather than an actual system outage. The tool provides a probability score for each event, indicating its potential impact on traffic, along with the affected geographies and the estimated number of people involved. This standalone prototype demonstrates how this functionality could be integrated into an AI SRE solution or monitoring tool.
主な機能
01Estimates the probability of an event impacting traffic.
02Supports multiple data sources via API integration for holidays and events.
03Leverages LLMs (GPT-4, Claude, Gemini, Deepseek) to identify external events.
04Can analyze traffic logs from a file.
05Identifies affected geographies and the number of people involved.
0616 GitHub stars
ユースケース
01Diagnosing traffic drops during holidays or major events.
02Integrating with monitoring tools for proactive alerting.
03Analyzing the impact of specific events on traffic patterns.