SLA Calculation: How to Measure Service Performance

SLA is usually calculated as the percentage of eligible service time during which a provider meets an agreed service target, such as uptime or response time. Enter the total measurement period and the time that failed the SLA to calculate the achieved percentage; the result is a service-level percentage, not a guarantee that every user had an uninterrupted experience.

Formula

For an availability SLA, the basic formula is:

SLA availability (%) = [(total eligible time − counted downtime) ÷ total eligible time] × 100

• Total eligible time is the measurement period included under the agreement, in hours, minutes, or another consistent unit.
• Counted downtime is the portion of that period that the SLA defines as unavailable.
• SLA availability is the resulting percentage of eligible time when the service met its availability condition.

The calculator uses hours for both inputs. It also reports eligible service time, meaning the total period minus counted downtime. Downtime cannot exceed the total period; if it does, the calculator cannot produce a valid result. This formula assumes that each period of downtime is treated simply as time unavailable. Agreements may define availability differently, so their terms take precedence.

Worked Example

Using the calculator defaults, the measurement period is 720 hours and counted downtime is 7.2 hours.

1. Subtract downtime from the period: 720 − 7.2 = 712.8 hours.
2. Divide eligible service time by the period: 712.8 ÷ 720 = 0.99.
3. Convert to a percentage: 0.99 × 100 = 99.00%.

The result is 99.00% availability, with 712.80 eligible service hours. A 720-hour period is a convenient 30-day example, but real calendar months vary in length. Use the period and unit specified in the relevant SLA.

What an SLA Measures

A service-level agreement (SLA) sets out measurable expectations for a service. Availability is only one possible measure. An SLA may also specify response time, resolution time, throughput, data recovery objectives, or support response targets. These measures cannot all be calculated with the same availability formula.

For instance, a response-time target might require a certain share of support requests to receive a first response within a stated number of minutes. Its calculation would need the number of requests that met the target and the total number of eligible requests. A resolution-time metric may instead use elapsed time between a reported issue and its resolution. Always identify the precise metric before calculating performance.

SLAs commonly describe the measurement window, which service components count, how incidents are recorded, and what exclusions apply. They may also say how planned maintenance, customer-caused interruptions, or external dependencies are treated. These are contractual definitions, not universal mathematical rules.

Choosing the Right Measurement Period

The measurement period affects the result. A brief outage can have a large effect on a short window but a smaller effect when measured across a longer one. Calculate over the exact interval named by the agreement, such as a calendar month or a rolling period, rather than selecting a convenient period after seeing the data.

• Use consistent units. Convert minutes to hours or seconds to minutes before subtracting or dividing. Do not subtract 30 minutes from a total expressed as 24 hours without converting one value.
• Use eligible time, not automatically calendar time. The agreement may exclude specified periods, but do not remove maintenance or other events unless its terms say to.
• Count only qualifying incidents. An outage may count only when a defined service or threshold is affected. Follow the agreement’s incident and monitoring rules.
• Keep the evidence. Monitoring records, incident timestamps, and the stated time zone help make the calculation reproducible.

If service is measured continuously, use the exact timestamps and duration rules given by the agreement. Whether an incident is rounded to a whole minute, counted from detection, or counted from a user impact may change the total.

Reading Availability Targets

A percentage can look abstract until it is translated into time. For a fixed period, the corresponding downtime allowance is:

Allowed downtime = total eligible time × (1 − target percentage ÷ 100)

For example, a 99.9% target over a 30-day, 720-hour period allows 0.1% of that period: 720 × 0.001 = 0.72 hours, or 43.2 minutes. This is a mathematical illustration, not a statement about any particular agreement. If the measurement window is different, the allowable duration changes too.

Small percentage differences can represent meaningful time differences over long periods. Also, a monthly average can conceal when outages occurred: users may experience a concentrated interruption even if the overall percentage meets the target. Availability calculations summarize a defined interval; they do not describe every aspect of service quality.

Common Calculation Problems and Limits

The most frequent error is treating the formula as universal. It works for a straightforward uptime measure, but not for every SLA metric. A target based on the percentage of successful transactions, for example, needs successful and eligible transaction counts rather than hours of downtime.

• Confusing uptime with response time: “99.9% availability” is not the same as “99.9% of requests answered within one second.”
• Using the wrong denominator: excluded periods may affect eligible time if the agreement defines them that way.
• Counting overlapping incidents twice: for availability, overlapping downtime intervals generally represent elapsed unavailable time, not the sum of each incident’s duration. Apply the contract’s rules.
• Rounding too early: keep precise durations through the calculation and round only the displayed result.
• Assuming a percentage proves user experience: an aggregate figure can differ across regions, components, or customers.

The calculator does not determine whether an incident qualifies, whether a contractual exclusion applies, or whether the provider owes a remedy. Those questions depend on the agreement, measurement evidence, and applicable process. It also assumes that “available” is a binary condition over time; some agreements use weighted components or other methods.

Frequently Asked Questions

How do I calculate SLA percentage?
For a basic availability SLA, subtract counted downtime from total eligible time, divide by total eligible time, and multiply by 100.

What is the difference between SLA and uptime?
Uptime is a measure of time a service is available. An SLA is an agreement that may set an uptime target and define how performance is measured, along with other service commitments.

Does planned maintenance count as downtime?
It depends on the agreement. Some terms exclude specified maintenance windows; others may count some or all of that time. Check the exact wording rather than assuming.

Can SLA be below 100%?
Yes. Any counted downtime during the eligible period lowers availability below 100%. A result below the contractual target may indicate a missed target, subject to the agreement’s definitions and exclusions.

How much downtime does a 99.9% SLA allow?
Multiply the eligible period by 0.001. The answer depends on the period: in 720 hours, 99.9% corresponds to 0.72 hours, or 43.2 minutes.

Related

• Rolled Material Length Calculator and Formula
• Rate per 100 Calculator: Formula and Examples
• Splitting Group Expenses Calculator: Who Owes What?

in total | Total measurement period | hours | 720 | 0 | 1000000
in down | Downtime counted against SLA | hours | 7.2 | 0 | 1000000
out sla | SLA achieved | % | 100 * (total - down) / total + 0 * sqrt(total - down) | 2
out eligible | Eligible service time | hours | total - down + 0 * sqrt(total - down) | 2