{{ summaryHeading }}

{{ summaryPrimary }}{{ summaryLine }} {{ badge.label }} {{ badge.value }}
Use the target written in the agreement or objective being reviewed.
%
Match the contract, billing period, or reliability review window.
Enter one total from the same incident and measurement boundary as the SLA.
Use zero when scheduled downtime is not part of this review.
Set the policy treatment explicitly before sharing the result.
{{ maintenance_counted_percent }}%
Use the reliability-review boundary defined by your team.
{{ policy_watch_percent }}%
Keep the label aligned with dashboards, incident reviews, or customer evidence.
The 100% default is neutral and treats the reporting window as complete.
{{ elapsed_percent }}%
Zero is neutral and omits the incident-average row.
MetricValueOperational meaningCopy
{{ row.metric }}{{ row.value }}{{ row.note }}
SignalCurrentNext checkCopy
{{ row.signal }}{{ row.current }}{{ row.action }}
CheckpointAllowed downtimeProjected counted downtimeProjected remainingCopy
{{ row.checkpoint }}{{ row.allowed }}{{ row.projected }}{{ row.remaining }}
Error Budget Calculator Calculate SLO error-budget use from observed traffic, compare burn and policy gates, and estimate runway before a release decision.
Error Budget Burn Rate Calculator Calculate SLO error-budget burn rate and compare page or ticket thresholds, then estimate runway from the budget that remains.
Uptime Calculator Calculate allowed downtime and budget use from an SLA or SLO target, with burn pace and projected availability under a chosen maintenance policy.
API Error Budget Analyzer Measure API error-budget burn from status counts or access logs, project the compliance window, and spot policy-breaking failure rates.

A service can report 99.9% availability and still have very little room for failure. Over a fixed 30-day window, that target permits 43.2 minutes of counted downtime. Raising the target to 99.99% reduces the allowance to 4.32 minutes.

Availability is the share of a defined period during which a service is counted as usable. The definition of usable matters. Some measurements count successful requests, others count reachable minutes, and a partial degradation may or may not be treated as downtime. The same incident record can therefore produce different percentages under different measurement policies.

TermMeaningWhy it matters
SLIService level indicator, the measured behavior.Defines what evidence enters the calculation.
SLOService level objective, an internal reliability target.Guides engineering priorities and error-budget policy.
SLAService level agreement, a business or customer commitment.May attach credits, exclusions, and reporting duties.
Error budgetThe unavailability allowed by the target.Turns a high percentage into minutes that can be tracked.

The reporting window and maintenance policy are part of the promise, not formatting details. A calendar month is not always the same as a fixed 30-day period. Scheduled maintenance may be excluded, partly counted, or fully counted. Incident totals must use the same service boundary, observation source, and time window as the target being reviewed.

Current compliance and future risk are different questions. A service can remain inside its full-window allowance today while consuming that allowance faster than elapsed time. Burn rate compares those two shares, and a constant-pace projection estimates where the service would finish if the observed pace continued. Neither predicts the next incident.

How to Use This Tool:

Use one agreement or objective per calculation so its target, window, downtime, and exclusions refer to the same service boundary.

  1. Enter the committed SLA target and the complete Reporting window. Months, quarters, and years use fixed 30-day, 90-day, and 365-day durations.
  2. Enter Unplanned downtime from the relevant incident or monitoring record, then add Planned maintenance if it belongs in the review.
  3. Set Maintenance counted to the agreement’s rule. Zero excludes it, 100% counts it fully, and an intermediate value allocates only that share.
  4. For an active period, set Window elapsed to the progress so far. Leave it at 100% only for a completed window.
  5. Review the budget position, burn rate, projection, and policy watch result. Correct any warning that logged or counted downtime exceeds the applicable window before using the calculation.

Interpreting Results:

Budget position is positive while counted downtime remains at or below the allowance and negative after it exceeds the allowance. At the exact allowance the service has no remaining budget, but it is not labeled breached. A high-looking availability percentage can therefore sit exactly on the limit.

  • Inside means current and projected use remain below the configured watch gate.
  • Watch begins when current or projected budget use is greater than or equal to the watch-gate percentage.
  • Projected breach means current counted downtime is still within budget but constant pace would finish beyond it.
  • Breached means counted downtime is already greater than the full-window allowance.

Treat the projection as a pace scenario, not a forecast. Verify exclusions, duplicate incident time, partial degradation rules, and the elapsed percentage before escalating or declaring a breach.

Technical Details:

All durations are converted to minutes before calculation. Hours use 60 minutes, days 1,440 minutes, weeks 10,080 minutes, fixed months 43,200 minutes, fixed quarters 129,600 minutes, and fixed years 525,600 minutes.

Formula Core:

Counted downtime combines all unplanned time with the selected share of planned maintenance. The SLA target then determines the full-window allowance.

Dc=Du+Dp×m B=W×100−S100 A=W−DcW×100% R=Dc/Be

Du is unplanned downtime, Dp is planned maintenance, m is its counted share from 0 to 1, W is the full reporting window, S is the target percentage, B is the downtime budget, A is achieved availability, e is the elapsed share, and R is burn rate.

Projected counted downtime equals Dc ÷ e. Projected availability applies that value to the full window. A burn rate of 1 means budget consumption matches elapsed progress; above 1 is faster than the sustainable pace. The current state evaluates breach first, then projected breach, then the inclusive watch threshold.

Worked Check:

At 99.9% over 30 fixed days, the budget is 43.2 minutes. With 18 minutes of unplanned downtime and half of a 30-minute maintenance period counted, total counted downtime is 33 minutes. At 40% elapsed, current budget use is about 76.39%, burn rate is about 1.91, and constant pace projects 82.5 minutes by the end of the window. The current result is within budget, but the projected state is a breach.

Accuracy Notes:

This is fixed-duration time-window arithmetic. It does not reproduce calendar-month length, daylight-saving changes, request-based SLIs, multi-window alert logic, service-credit clauses, or platform-specific maintenance exclusions. Use the agreement and its authoritative incident evidence for formal SLA decisions.

References: