Explore how Google's Serverless SRE team tackles platform-level latency issues in this 47-minute conference talk from YOW! Australia 2023. Dive into detecting regressions, measuring their impact, and tracking performance over time. Learn about the limitations of SLOs and discover a statistical approach for a customer-centric view of platform performance. Gain insights on total latency distribution, request delivery latency, reliability in practice, and the 2-Sigma Technique. Understand concepts like stationarity, overload score, and impact analysis. Examine the mechanics, FAQ, backtesting, limitations, and other applications of this approach. Conclude with valuable lessons for platform engineers and SREs working on serverless architectures.
Why Is My App Slow - Defining Reliability in Platform Engineering