Course contentsShow
← Cloud and Infrastructure
Lesson 2117 of 2,98945. Observability - Dashboards and SLOsFree lesson

Setting Realistic SLO Targets

Understand how to choose appropriate SLO thresholds based on user expectations and business requirements.

Setting Realistic SLO Targets

What you'll learn: How to choose appropriate SLO thresholds that balance user happiness, technical feasibility, and business constraints.

Why SLO Targets Must Be Realistic

You've defined your SLIs (the what you measure) and their specifications (the how). Now comes the crucial decision: setting the target threshold — the percentage or value that defines success. Set it too high, and your team burns out chasing unattainable perfection. Set it too low, and users suffer poor experiences while you're technically "meeting" your goals.

The Balancing Act

Think of SLO targets like thermostat settings. You want your home comfortable, but setting it to a perfect 72°F year-round at any cost would be prohibitively expensive and unnecessary when 68-74°F would satisfy everyone.

User Expectations First

Start by understanding what users actually need, not what engineers think is "good enough." A payment API might require 99.9% availability, but a nightly analytics report processor might work fine at 95%. Ask: At what point does degradation cause user pain?

Business Requirements

Factor in contractual obligations (SLAs), competitive positioning, and revenue impact. Your SLO should support the SLA with a comfortable buffer — if you promise 99.9% uptime in contracts, target 99.95% internally.

Technical Reality

Consider your current performance baseline, infrastructure complexity, and operational maturity. Jumping from 95% to 99.99% overnight is unrealistic. Start where you are, trend upward deliberately.

Error Budgets

Remember that 100 - SLO gives you an error budget. A 99.9% target means 0.1% allowable failure — that's your innovation budget for risky deployments and experimentation.

Key Takeaway: Realistic SLO targets emerge from the intersection of user needs, business constraints, and technical capability — not arbitrary "nines" or wishful thinking. Start conservatively, measure honestly, and adjust based on evidence.