SRE Thailand Bootcamp

Week 5: SLOs, SLIs, Error Budgets

Theme: Define and measure service reliability quantitatively.

Learning Objectives

1. Define SLIs (latency, availability, throughput, durability) from metrics
2. Set SLO targets based on user expectations
3. Calculate error budget and error budget burn rate
4. Implement burn-rate alerts in Prometheus

Reading Materials

• Google SRE Book - Chapter 4: Service Level Objectives
• Google SRE Workbook - Chapter 5: Alerting on SLOs
• SLOs with Prometheus (Robust Perception)
• Error Budget Burn Rate Alerts (Grafana)
• Sloth - SLO Generator

Finger Exercises (choose 2)

1. Install sloth and generate SLO rules for the Go app.
2. Write a PromQL query to calculate error budget remaining for a 99.9% SLO over 30 days.
3. Set up a burn-rate alert using the multi-window ratio technique with fast (5m) and slow (1h) burn rate windows.

Project

Define 3 SLOs for the Phase 1 app:

• Latency SLO: 99% of requests < 500ms over 30 days
• Availability SLO: 99.9% of requests succeed over 30 days
• Throughput SLO: 95% of orders processed within 5s over 7 days

Implement burn-rate alerts using Prometheus recording rules (or sloth). Create a Grafana "SLO Dashboard" showing:

• SLI values over time
• Error budget remaining
• Burn rate (fast/slow)
• SLO status (compliant / at risk / burning)

Write a brief document explaining each SLO choice and the user journey it maps to.

Time Budget