Understanding and Calculating Uptime SLAs: What 99.9% Really Means

The Availability Formula: How to Calculate Uptime

Calculating availability follows a simple formula: A = (T - D) / T x 100%. Where T is the total time of the measurement period and D is the total downtime within that period. A month has 43,200 minutes (30 days) or 44,640 minutes (31 days). A year has 525,600 minutes.

Example calculation: your website was unreachable for a total of 45 minutes in March (31 days = 44,640 minutes). The availability is: (44,640 - 45) / 44,640 x 100 = 99.899%. That falls just below a 99.9% SLA. A 45-minute outage in one month already means an SLA violation for a 99.9% SLA.

The Nines Scale: From 99% to 99.999%

In the industry, availability levels are referred to as nines. Each additional nine dramatically reduces the allowed downtime and correspondingly increases the requirements for infrastructure and processes.

What Should Be in an SLA: The Key Clauses

An SLA is more than a percentage. The details of the agreement determine whether the SLA actually provides protection or is just a marketing statement. Pay attention to the following points when reviewing or negotiating an SLA.

Definition of Downtime

The definition of downtime is the most critical point. Some providers count only complete server failures as downtime. Partial outages, performance degradation, or reachability problems from specific regions are not considered. A good SLA defines downtime as any interruption that prevents access to the service or increases response time above a defined threshold.

Measurement Method and Period

How is availability measured? By the provider itself or by an independent service? At what intervals (every minute, every 5 minutes, hourly)? An SLA measured hourly misses short outages under one hour. The measurement period (monthly, quarterly, annually) affects the allowed downtime per period.

Exceptions and Maintenance Windows

Most SLAs exclude planned maintenance from the calculation. This is legitimate when maintenance windows are clearly defined and announced. It becomes problematic when the provider claims unlimited maintenance windows or retroactively declares downtime as maintenance to avoid SLA violations.

Compensation for SLA Violations

What happens when the provider fails to meet the SLA? Typical compensations are service credits (credits on the next invoice) between 5 and 30 percent of the monthly fee. Some SLAs offer tiered compensation: 10% for missing by 0.1 percentage points, 25% for 0.5 percentage points, and so on.

Planned vs. Unplanned Downtime

The distinction between planned and unplanned downtime is crucial for SLA evaluation. Planned maintenance (e.g., server updates on Sunday morning) is excluded from calculation in most SLAs. Unplanned outages (server crash, network issue, DDoS attack) count as downtime.

From the customer's perspective, all downtime is relevant, whether planned or not. Your visitors don't know whether the website is offline due to maintenance or a problem. Therefore, you should include planned maintenance in your availability calculation and prefer providers that offer rolling updates without downtime.

Verifying SLA Compliance with Monitoring

Without independent monitoring, you cannot verify your provider's SLA compliance. You rely on the provider's data, and they naturally have an interest in reporting availability as high as possible. External monitoring provides independent data that can serve as the basis for SLA claims.

Monitoring Requirements

• Minute-accurate checks: Measurement intervals of maximum 1 minute to capture even short outages. A 5-minute interval misses outages of 1 to 4 minutes
• Multi-location tests: Checks from multiple locations to detect regional outages and rule out false alarms from local network issues
• Logging: Complete recording of all checks with timestamp, response time, and HTTP status code as evidence for SLA claims
• Automatic calculation: Automatic availability calculation per day, week, month, and year with configurable SLA thresholds
• Outage reports: Detailed reports for each outage with start time, duration, and affected locations

XICTRAQ as SLA Monitoring Tool

XICTRAQ meets all these requirements. The uptime monitoring function checks your website at configurable intervals from multiple locations and automatically calculates actual availability. Outages are logged to the minute and presented in clear reports.

Particularly valuable for SLA monitoring: XICTRAQ distinguishes between complete outages (HTTP 5xx, connection timeout) and performance degradation (response time above threshold). This lets you also verify SLAs that include performance requirements, not just pure reachability.

SLA Budget Tracking: Remaining Downtime at a Glance

A useful concept in SLA management is the SLA budget (also called error budget). It shows how much downtime remains within the current measurement period before the SLA is violated. For a 99.9% SLA and a month with 43,200 minutes, the budget is 43.2 minutes. If 20 minutes of downtime have already been used, 23.2 minutes remain.

XICTRAQ calculates the remaining SLA budget in real time. When the budget is running low, you receive a warning before the SLA is actually violated. This gives you time to act proactively: postpone maintenance, contact the hosting provider, or activate redundancy measures.

Typical SLA Traps: What to Watch For

SLA Levels by Industry: What Is Common?

Availability requirements vary significantly by industry and use case. The following values are practical benchmarks.

Increasing Availability: Measures by SLA Level

For 99.9% (Three Nines)

• Quality hosting with automatic restart on server failure
• Uptime monitoring with minute-accurate checks and immediate notification
• Automated backups with tested restoration
• CDN deployment for static resources to offload the origin server

For 99.99% (Four Nines)

• Everything from the 99.9% level plus:
• Redundant servers with load balancing and automatic failover
• Redundant database with replication (primary-replica or multi-primary)
• Blue-green or canary deployments for zero-downtime updates
• Multi-location monitoring from at least three locations

For 99.999% (Five Nines)

• Everything from the 99.99% level plus:
• Geo-redundant infrastructure in at least two data centers
• Automatic DNS failover between data centers in under 30 seconds
• Chaos engineering practice (regular failure simulations)
• 24/7 on-call with defined response time under 5 minutes

Mean Time to Recovery (MTTR) as Second Key Metric

Beyond availability, Mean Time to Recovery (MTTR) is a crucial metric. It measures how quickly a service is restored after an outage. A low MTTR partially compensates for higher outage frequency: when outages are resolved quickly, total downtime remains low.

MTTR consists of four phases: detection (how quickly is the outage noticed?), diagnosis (what is the cause?), repair (how is the problem solved?), and verification (is the service stable again?). Monitoring reduces the detection phase to minutes instead of hours. XICTRAQ's alerting function ensures the right people are immediately notified, accelerating the diagnosis phase.

SLA Reporting with XICTRAQ

XICTRAQ offers comprehensive SLA reports suitable for both internal analysis and communication with hosting providers.

• Availability reports: Automatically calculated availability per day, week, month, and year with comparison to SLA targets
• Outage logs: Detailed list of all outages with start time, end time, duration, and HTTP status code
• Performance trends: Long-term development of response times that makes gradual degradation visible
• Status page: Public or internal status page showing current and historical availability status
• SLA budget tracker: Remaining downtime budget in real time with warning when running low

Start now with a free demo and see your actual availability at a glance. Or register to continuously monitor and document SLA compliance.