Course contentsShow
← Cloud and Infrastructure
Lesson 2492 of 2,98953. High Availability PatternsFree lesson

Steady State Hypothesis

Defining steady state metrics and establishing baseline system behavior before introducing chaos experiments.

Steady State Hypothesis

What You'll Learn

You'll learn how to define measurable steady state metrics and establish baseline system behavior before running chaos experiments.

Understanding Steady State

Before you intentionally break things with chaos engineering, you need to know what "normal" looks like. The Steady State Hypothesis is your definition of how a healthy system behaves under typical conditions. It's not about perfection—it's about capturing the expected behavior you want to verify remains stable during chaos experiments.

Think of it like a pilot's pre-flight checklist. You verify all instruments read "normal" before takeoff, so you can immediately detect if something goes wrong mid-flight.

Core Components

Your steady state hypothesis consists of:

Observable Metrics: Define concrete, measurable indicators of system health. Examples include:

  • Response time percentiles (p95, p99)
  • Error rates (below 0.1%)
  • Throughput (requests per second)
  • Resource utilization (CPU, memory)
  • Active connections

Baseline Values: Establish acceptable ranges from historical data. For instance: "API response time stays below 200ms for 99% of requests during business hours."

Measurement Window: Specify how long you observe to confirm steady state—typically 5-15 minutes before introducing chaos.

Why This Matters

Without a steady state hypothesis, you can't prove chaos experiments cause problems—or that your system recovers properly. You're essentially guessing. The hypothesis transforms chaos engineering from random breaking into scientific experimentation with falsifiable predictions.

Key Takeaway: The Steady State Hypothesis defines measurable "normal" system behavior using specific metrics and acceptable ranges, establishing the baseline you'll verify remains stable during chaos experiments.