What Is an SLA? Uptime Percentages Explained (99.9% vs 99.99%)
What does 99.9% uptime really mean? Learn how SLA uptime percentages translate to real downtime — from 3.65 days/year (99%) to 5 minutes/year (99.999%). Includes calculator.
Your hosting provider promises 99.9% uptime. Your SaaS vendor guarantees 99.95%. Your cloud platform advertises 99.99%. These numbers look nearly identical — they’re all above 99%. But the real-world difference between them is enormous.
Understanding what these percentages actually mean is essential for evaluating vendor promises, setting expectations for your own services, and writing SLA agreements that protect your business.
What Is an SLA?
A Service Level Agreement (SLA) is a formal commitment between a service provider and a customer that defines the expected level of service. In the context of web hosting, cloud services, and SaaS, the most critical metric in an SLA is usually uptime — the percentage of time the service is guaranteed to be available.
An SLA typically includes three components: the uptime guarantee (expressed as a percentage), the measurement period (usually monthly), and the remedies for failure (typically service credits or refunds).
For example, an SLA might state: “We guarantee 99.9% uptime per calendar month. If uptime falls below this threshold, affected customers receive a service credit equal to 10% of their monthly fee for each 0.1% below the guarantee.”
The Uptime Table: What the Numbers Really Mean
This is where the decimals matter. Each additional “9” in the percentage dramatically reduces the acceptable downtime.
| Uptime % | Called | Downtime/Year | Downtime/Month | Downtime/Week |
|---|---|---|---|---|
| 99% | Two nines | 3.65 days | 7.31 hours | 1.68 hours |
| 99.5% | — | 1.83 days | 3.65 hours | 50.4 minutes |
| 99.9% | Three nines | 8.77 hours | 43.8 minutes | 10.1 minutes |
| 99.95% | Three and a half nines | 4.38 hours | 21.9 minutes | 5.04 minutes |
| 99.99% | Four nines | 52.6 minutes | 4.38 minutes | 1.01 minutes |
| 99.999% | Five nines | 5.26 minutes | 26.3 seconds | 6.05 seconds |
The jump from 99.9% to 99.99% looks like a tiny improvement — just 0.09 percentage points. But in real terms, it’s the difference between nearly 9 hours of acceptable downtime per year and less than 53 minutes. That’s a 10x improvement in required reliability.
What Most Providers Actually Offer
Here’s a rough breakdown of what’s standard at different tiers:
99% uptime is unusually low for any professional service. It allows over 3.5 days of downtime per year. Most shared hosting plans have historically targeted this level, though many have moved higher.
99.9% uptime is the most common SLA tier for SaaS products, managed hosting, and standard cloud services. It allows about 8.7 hours of annual downtime — roughly one significant outage per year.
99.95% uptime is offered by premium SaaS and cloud providers as a step up. It cuts acceptable downtime to about 4.4 hours per year.
99.99% uptime is the standard for major cloud infrastructure providers like AWS, Google Cloud, and Azure for their core services. Achieving this requires redundant systems, automated failover, and sophisticated monitoring. The allowed downtime is under an hour per year.
99.999% uptime (five nines) is the gold standard, allowing only about 5 minutes of downtime per year. Very few services achieve this consistently. It typically requires geographically distributed redundancy with zero-downtime deployments.
How Uptime Is Measured
Not all uptime measurements are created equal. When evaluating an SLA, pay attention to how the provider defines and measures availability.
Measurement window — Most SLAs use monthly measurement periods. A provider with 99.9% monthly uptime could have a bad week (an hour of downtime) and still meet the SLA for the month.
Exclusions — Many SLAs exclude scheduled maintenance windows from uptime calculations. A provider that performs 4 hours of monthly maintenance and guarantees 99.9% uptime outside those windows has effectively lower real-world uptime than the number suggests.
Monitoring frequency — How often is uptime checked? If a provider measures availability every 5 minutes, a 3-minute outage might not register at all. The granularity of monitoring affects the accuracy of uptime reporting.
Definition of “down” — Some providers define downtime as a complete failure to respond. Others include degraded performance (slow response times) in their downtime calculations. Read the fine print.
SLA Credits: Are They Worth Anything?
When providers breach their SLA, the standard remedy is a service credit — a percentage discount on your next bill. Typical structures:
- Below 99.9% but above 99.0%: 10% credit
- Below 99.0% but above 95.0%: 25% credit
- Below 95.0%: 50% credit (sometimes 100%)
Here’s the uncomfortable truth: SLA credits rarely compensate you for the actual cost of downtime. If you’re paying $100/month for hosting and an outage costs your business $5,000 in lost revenue, a $10 service credit doesn’t come close. SLA credits are a signal of commitment, not insurance against loss.
For mission-critical services, the SLA matters less than the provider’s actual track record. Look at their historical uptime data (often published on their status page) rather than relying solely on the contractual guarantee.
Setting Your Own SLA
If you’re a service provider defining your own SLA, here’s how to think about it:
Start with your actual reliability. Measure your real uptime for 3-6 months before committing to an SLA number. Set the guarantee below your actual performance — you need headroom for unexpected incidents.
Factor in dependencies. Your uptime can’t exceed the uptime of your infrastructure dependencies. If you run on a cloud provider with a 99.99% SLA, your service can’t credibly guarantee 99.999%.
Define maintenance windows clearly. If you need regular maintenance downtime, exclude it from the SLA but commit to advance notice (24-72 hours is standard).
Make credits meaningful but sustainable. Credits should be large enough that customers feel the commitment is real, but not so large that a single bad month threatens your business.
Monitor continuously. You can’t manage an SLA you don’t measure. Set up monitoring from multiple locations, track response times alongside availability, and build automated SLA reporting.
Monitoring as the Foundation
Every SLA depends on accurate measurement. You can’t prove compliance — or catch breaches — without continuous monitoring from external infrastructure.
The monitoring check interval matters for SLA calculation. If your SLA promises 99.99% uptime (52.6 minutes/year), a monitoring tool that checks every 30 minutes could miss a 25-minute outage entirely if the checks happen to land before and after the incident.
For SLA-grade monitoring, use 1-minute check intervals from multiple geographic locations. This gives you high-resolution data that accurately reflects your true availability.
Keep monitoring logs as documentation. When customers dispute uptime or when you need to claim credits from your own vendors, historical monitoring data is your evidence.
Monitor your SLA compliance with OpsKitty — track uptime from 29 global regions with checks as frequent as every 30 seconds. Start free today.