Uptime Math
SLA Uptime Calculator
How Much Downtime Does 99.9%, 99.99%, 99.999% Allow?
Updated April 2026
Calculator Inputs
Downtime Allowances at 99.900% Uptime
Per Day
1.44 min
Per Week
10.08 min
Per Month
43.80 min
Per Quarter
2.19 hr
Per Year
8.76 hr
Annual Exposure at This SLA Tier
$2,628,000
If you have 526 minutes of downtime per year and each hour costs you $300,000
Complete SLA Downtime Allowance Table
| Uptime % | Nines | Per Year | Per Month | Per Week | Per Day |
|---|---|---|---|---|---|
| 90% | 1 nine | 52,560 min (36.5 days) | 4,380 min | 1,008 min | 144 min |
| 95% | 1.3 nines | 26,280 min (18.25 days) | 2,190 min | 504 min | 72 min |
| 99% | 2 nines | 5,256 min (3.65 days) | 438 min | 100.8 min | 14.4 min |
| 99.5% | 2.3 nines | 2,628 min (1.82 days) | 219 min | 50.4 min | 7.2 min |
| 99.9% | 3 nines | 525.6 min (8h 46m) | 43.8 min | 10.08 min | 1.44 min |
| 99.95% | 3.3 nines | 262.8 min (4h 23m) | 21.9 min | 5.04 min | 0.72 min |
| 99.99% | 4 nines | 52.56 min | 4.38 min | 1.01 min | 8.64 sec |
| 99.999% | 5 nines | 5.26 min | 26.3 sec | 6.05 sec | 0.86 sec |
| 99.9999% | 6 nines | 31.5 sec | 2.63 sec | 0.60 sec | 0.086 sec |
How to Calculate Downtime From an Uptime Percentage
An SLA percentage is a promise about uptime. The allowed downtime is simply the remaining fraction multiplied by the length of the period:
The only constants you need are the minutes in each period: a year is 525,600 minutes (365 × 24 × 60) and a day is 1,440 minutes.
Worked example: 99.9% uptime
- Downtime fraction: 1 − 0.999 = 0.001 (0.1%)
- Per year: 0.001 × 525,600 = 525.6 minutes (about 8 hours 46 minutes)
- Per day: 0.001 × 1,440 = 1.44 minutes
Add a nine and the allowance shrinks by 10×: 99.99% is 0.0001 × 525,600 = 52.56 minutes per year, and 99.999% is 5.26 minutes per year. The table above applies this same formula across every tier from 99.5% to 99.9999%, per year, month, week, and day.
What Each Tier Means in Practice
99.9% (Three Nines)
About 8 hours 46 minutes of downtime per year. This sounds good until you realize that 8 hours could be one single Saturday afternoon outage. For most production systems, this means: single-region deployment with solid monitoring and manual escalation. Most shared hosting and basic cloud deployments fall into this tier naturally.
99.99% (Four Nines)
Only 52 minutes of downtime per year. That is less than one hour, spread across all incidents. This requires genuinely automated failover - you cannot manually respond to every incident in under 5 minutes reliably. Multi-AZ with database replication and load balancer health checks is the minimum architecture. This is the standard for most SaaS products with paying B2B customers.
99.999% (Five Nines)
5 minutes 16 seconds per year. Effectively zero planned maintenance is possible at this tier. Multi-region active-active architecture is required. This is what financial exchanges, telephone networks, and critical healthcare systems target. The cost delta from four nines to five nines is typically 10-20x in infrastructure spend.
SLA Tier to Architecture Map
| SLA Tier | Architecture Required | Additional Cost Est. |
|---|---|---|
| 99.9% | Single region, multi-AZ with monitoring | ~$5K-20K/yr extra |
| 99.99% | Multi-AZ with automated failover, DB replication, load balancing | ~$50K-200K/yr extra |
| 99.999% | Multi-region active-active, global load balancing, chaos engineering | ~$500K-2M/yr extra |
Architecture costs are rough estimates for typical cloud deployments on AWS/Azure/GCP. Actual costs depend on scale, region count, and data volume.