Course contentsShow
← Software Engineering
Lesson 1176 of 1,93335. Monitoring and AlertingFree lesson

Calculating and Tracking Error Budgets

Learn practical methods for computing remaining error budget and using it to inform operational decisions.

Calculating and Tracking Error Budgets

You'll learn how to turn error budgets from abstract targets into concrete operational metrics that guide release and incident response decisions.

What Is Error Budget Calculation?

An error budget is your allowed failure time. If your SLO promises 99.9% availability, you have a 0.1% error budget—roughly 43 minutes of downtime per month. Calculating and tracking this budget transforms reliability from philosophy into measurable practice.

The basic formula is straightforward:

  • Error budget (time) = (1 - SLO) × time period
  • For 99.9% monthly: (1 - 0.999) × 30 days = 43.2 minutes

Tracking Consumption

Track your budget by measuring actual failures against allowed failures:

  1. Count bad events: Failed requests, slow responses beyond threshold, or downtime minutes
  2. Calculate consumption rate: (bad events / total events) × 100%
  3. Compare to budget: Have you used 10%, 50%, or 90% of your allowed failures?

If your SLO allows 0.1% failure and you're currently at 0.05% failure rate, you've consumed 50% of your error budget.

Using Budget to Drive Decisions

This is where error budgets become powerful:

  • Budget remaining? You can take calculated risks: deploy on Friday, test new features in production, experiment with infrastructure changes
  • Budget exhausted? Freeze risky changes, focus exclusively on reliability improvements, defer feature work until health improves
  • Trending toward depletion? Increase review rigor, add monitoring, plan reliability sprints

Think of it like a bank account: when you have savings, you can invest; when you're overdrawn, you must conserve.

Key Takeaway: Error budgets become actionable when you calculate remaining budget regularly, track consumption trends over time, and use these metrics to balance feature velocity against reliability needs—making reliability engineering objective rather than reactive.