Error Budget Calculator
Calculate and understand your service's error budget
What is an Error Budget?
An error budget is the acceptable amount of downtime your service can have while still meeting your Service Level Objective (SLO). It answers the question: "How much failure can we afford?"
Key Concept
- SLO = 99.9% means you can afford 43.2 minutes of downtime in a 30-day window
- This is your "budget" to spend on experiments, deployments, and maintenance
- When you've used up your budget, focus on stability instead of new features
Understanding the Calculator
Main Components
1. SLO Target Selection
Choose from preset SLO targets:
- 99% - Large consumer services (okay with ~7.2 hours of downtime per 30 days)
- 99.5% - Most enterprise services
- 99.9% - Critical business services
- 99.95% - Very critical services
- 99.99% - Ultra-critical infrastructure
2. Key Metrics Displayed
| Metric | What It Means | Use Case |
|---|---|---|
| Allowed Downtime | Minutes you can be down in a 30-day window | Budget planning |
| In Hours | Same metric in hours | Planning maintenance windows |
| Failure Rate | Share of requests (or time) allowed to fail | Understanding error rate |
3. Progress Bar
Visual representation of your SLO target (100% = perfect uptime)
4. Burn Rate Check
Enter the downtime so far and how many days into the 30-day window you are. The calculator compares how fast you are spending the budget with the pace that would make it last the whole window.
Burn Rate: The Critical Metric
Burn rate is how fast you are spending the error budget, compared with the pace that would use exactly all of it by the end of the window. At 1× you finish the window exactly on target. At 2× the budget is gone halfway through.
Burn Rate Categories
| Burn Rate | What It Means | Action |
|---|---|---|
| 1× or less | On pace: the budget lasts the whole window | Spend it on releases and experiments |
| Above 1× | Above pace: the budget runs out before the window ends | Find out what is consuming it |
| 2× or more | Fast burn: the budget is gone by mid-window or sooner | Alert, and slow risky releases |
| Budget used up | Exhausted: the SLO is missed for this window | Restrict or stop releases until back within SLO |
Sources: the 1× baseline and the 2× alert come from Google Cloud's Learn how to set SLOs. Slowing releases for an endangered SLO comes from How we build good SLOs at Google, and restricting them once the SLO is missed from What are the consequences of SLO violations.
How to Calculate Burn Rate
or, equivalently: Observed Error Rate ÷ Allowed Error Rate (100% − SLO)
Example:
- Your SLO: 99.9% = 43.2 minutes of budget per 30 days
- Day 10, with 5 minutes of downtime so far
- Budget used: 5 / 43.2 = 11.6%. Window elapsed: 10 / 30 = 33.3%
- Burn Rate: 11.6 / 33.3 = 0.35× (on pace)
- The same 5 minutes by day 1 would be 11.6 / 3.3 = 3.5×, and the budget would run out around day 9
The calculator averages the burn since the window started. Alerting also looks at shorter windows: Google's guidance is to page when a large amount of budget burns over a short window, and to open a ticket when the burn rate stays elevated over a longer one.
Using Error Budgets for Deployment Decisions
Decision Framework
Is the error budget used up?
├─ YES → Restrict or stop risky releases until back within SLO
└─ NO → Is the burn rate above 1×?
├─ NO → On pace: deploy and experiment
└─ YES → Find the cause; at 2× or more, slow risky releases
Practical Examples
Scenario 1: On Pace
- SLO: 99.9% (43.2 min / 30 days)
- So far: 10 minutes of downtime by day 15
- Burn Rate: 23.1% used ÷ 50% elapsed = 0.46×
- Decision: Keep shipping; the budget lasts the window
Scenario 2: Budget Exhausted
- SLO: 99.9% (43.2 min / 30 days)
- So far: 45 minutes of downtime by day 20
- Budget used: 104%, so the SLO is missed for this window
- Decision: Restrict or stop risky releases until back within SLO
Scenario 3: Fast Burn
- SLO: 99.99% (4.32 min / 30 days)
- So far: 1.5 minutes of downtime by day 3
- Burn Rate: 34.7% used ÷ 10% elapsed = 3.5×
- Decision: At this pace the budget is gone around day 9; slow risky releases and find the cause
The Math Behind It
Calculate Downtime from SLO
Budget Minutes = (Error Rate / 100) × Window Minutes
30-day window = 30 × 24 × 60 = 43,200 minutes
Example: 99.9% SLO
- Error Rate: 100 - 99.9 = 0.1%
- Budget: (0.1 / 100) × 43,200 = 43.2 minutes per 30 days
Which Window?
Always say which window a budget covers. This site uses a rolling 30-day window. Some tools use 28 days or a calendar month, and an average month (one twelfth of a year, 730 hours) gives 43.8 minutes for the same 99.9% target. Google's SLO guidance notes that rolling windows are more closely aligned with user experience, while calendar windows line up with business targets and planning.
Source: Google Cloud, Learn how to set SLOs.
Common SLO Reference Table
| SLO | Per Year | Per 30 Days | Per Week | Per Day |
|---|---|---|---|---|
| 99% | 3.65 days | 7.2 hours | 1.68 hours | 14.4 min |
| 99.5% | 1.83 days | 3.6 hours | 50.4 min | 7.2 min |
| 99.9% | 8.76 hours | 43.2 min | 10.1 min | 1.44 min |
| 99.95% | 4.38 hours | 21.6 min | 5.04 min | 43.2 sec |
| 99.99% | 52.6 min | 4.32 min | 60.5 sec | 8.64 sec |
| 99.999% | 5.26 min | 25.9 sec | 6.05 sec | 0.86 sec |
Best Practices
DO
- Track burn rate daily - Make it visible to the team
- Use budget wisely - Spend on meaningful improvements
- Reset monthly - Start fresh each month
- Communicate openly - Share burn rate status with stakeholders
- Link to deployment decisions - Let burn rate guide your deployment cadence
DON'T
- Ignore your budget - It exists for a reason
- Save the entire budget - You're leaving productivity on the table
- Wait until 100% burned - Start focusing on stability at ~70%
- Blame ops for missing SLO - It's a team responsibility
- Set SLO too high - Be realistic about your infrastructure
Advanced Usage
Multi-Service Strategy
If you have multiple services:
| Service | SLO | Budget / 30 Days | Budget per Day at 1× |
|---|---|---|---|
| API | 99.99% | 4.32 min | 8.64 sec |
| Web UI | 99.9% | 43.2 min | 1.44 min |
| Background Jobs | 99% | 7.2 hrs | 14.4 min |
Monthly Planning
Week 1: Review last month's burn
- Did we miss SLO? Why?
- What caused outages?
- Start month fresh
Week 2-3: Aggressive deployment phase
- Burn rate is low
- Deploy new features safely
- Run experiments
Week 4: Stability phase
- Burn rate is rising
- Focus on bug fixes
- Stabilize system
Implementation Checklist
Related Concepts
SLO (Service Level Objective)
The target (e.g., 99.9%)
SLI (Service Level Indicator)
The measurement (e.g., request success rate)
SLA (Service Level Agreement)
The contract (e.g., refund if we miss SLO)
MTTD (Mean Time To Detection)
How fast we notice problems
MTTM (Mean Time To Mitigation)
How fast user impact stops
MTTR (Mean Time To Repair)
How long the full fix takes
Common Questions
Q: Should we always target 99.99%?
A: No. Higher SLO = more operational burden. Target what your customers need, not what's theoretically possible.
Q: What if we never hit our SLO?
A: Either your SLO is too strict, or you need more engineering effort. Consider lowering the SLO or investing in reliability.
Q: Can we use error budget to skip incidents?
A: No. Incidents still happen. Error budget lets you decide when to deploy risky changes, not when to ignore problems.
Q: How do we track burn rate?
A: Divide the share of the budget used by the share of the window elapsed (or the observed error rate by the allowed error rate). Display it on dashboards, Slack, or status pages.
Q: What's a "good" burn rate?
A: Around 1× or below over the window. You meet the SLO and still use the budget for releases and experiments. Staying far below 1× for a long time can mean the target is stricter than users need, which means you are spending more on reliability than you need to.