Course contentsShow
← System Design
Lesson 1278 of 1,91932. Monitoring and AlertingFree lesson

Multiple SLOs for Comprehensive Coverage

Understand why systems need multiple SLOs covering different aspects: availability, latency percentiles, and error rates.

Multiple SLOs for Comprehensive Coverage

What you'll learn: Why a single SLO isn't enough and how multiple SLOs work together to give complete visibility into service health.

The Problem with One-Size-Fits-All

A single SLO is like judging a restaurant only by how fast it serves food. Sure, speed matters—but what about food quality, order accuracy, and whether the kitchen is on fire? Similarly, measuring only availability (like "99.9% uptime") misses critical dimensions of user experience.

Why Multiple SLOs?

Different aspects of your service affect users in different ways, and a single metric can't capture everything:

Availability SLO tells you if your service responds at all—but not how well. A service might be "up" yet returning errors or taking 30 seconds per request.

Latency SLO measures how fast requests complete. Users care about speed, and a slow service feels broken even if it's technically available. You typically track percentiles (like p95 or p99) to catch tail latencies that affect real users.

Error Rate SLO tracks the percentage of requests that fail. Even with 100% availability and fast responses, a 5% error rate means one in twenty users has a bad experience.

Comprehensive Coverage in Action

Imagine your API has these SLOs:

  • Availability: 99.9% of health checks succeed
  • Latency: 95% of requests complete under 200ms
  • Error rate: 99.5% of requests return non-error status codes

Together, these paint a complete picture. If availability is fine but latency degrades, you know the system is up but struggling. If errors spike while availability looks good, you're serving responses—just broken ones.

The Right Combination

Choose SLOs that align with what users actually experience. Most services need at least two or three covering different dimensions. This prevents blind spots where technical metrics look healthy while users suffer.

Key Takeaway: Multiple SLOs covering availability, latency, and error rates provide comprehensive visibility—no single metric can capture the full user experience.