SLOs, SLIs and Error Budgets

Service Level Objectives (SLOs) provide a framework for defining and measuring reliability from the user's perspective. Rather than chasing 100% uptime (which is impossible and infinitely expensive), SLOs set explicit reliability targets that balance user happiness with engineering velocity.

---

Key Definitions

Term	Definition	Example
SLI (Service Level Indicator)	A quantitative measure of service behaviour	99.2% of requests complete in under 500ms
SLO (Service Level Objective)	A target value or range for an SLI	99.5% of requests should complete in under 500ms
SLA (Service Level Agreement)	A contract with consequences for missing an SLO	"99.9% uptime or we refund 10% of your bill"
Error Budget	The allowed amount of unreliability (1 - SLO)	0.5% of requests can fail over 30 days

The Relationship

SLI (measurement) → SLO (target) → SLA (contract) → Error Budget (remaining allowance)

• SLIs tell you how you are performing
• SLOs tell you how well you need to perform
• SLAs are business commitments with financial or contractual penalties
• Error budgets tell you how much room you have left

---

Choosing Good SLIs

Good SLIs measure what users actually experience:

Common SLI Types

SLI Type	Metric	Good Event
Availability	Successful requests / total requests	Non-5xx response
Latency	Requests faster than threshold / total requests	Response in < 500ms
Correctness	Correct responses / total responses	Response matches expected output
Throughput	Successful operations / time	Orders processed per minute
Freshness	Data updated within threshold / total data	Dashboard refreshed in < 60s

SLI Best Practices

• Measure at the edge — as close to the user as possible (load balancer, CDN)
• Use ratios — "good events / total events" is the standard SLI format
• Keep it simple — 3-5 SLIs per service is usually sufficient
• Avoid vanity metrics — "CPU usage < 80%" is not an SLI (users do not care about CPU)

---

Setting SLOs

Choosing a Target

SLO Target	Monthly Error Budget	Monthly Downtime Equivalent
99%	1%	~7.3 hours
99.5%	0.5%	~3.6 hours
99.9%	0.1%	~43.8 minutes
99.95%	0.05%	~21.9 minutes
99.99%	0.01%	~4.4 minutes

Guidelines for Target Selection

• Start with current performance — if your service is currently at 99.7%, do not set a 99.99% SLO
• Align with user expectations — an internal tool has different needs than a payment API
• Leave room for error budgets — too-tight SLOs stifle feature development
• Review quarterly — adjust as the service matures and user needs change

Tip: "Nines" beyond 99.9% are exponentially more expensive to achieve. Each additional nine roughly costs 10x more.

---

Error Budgets