Turn an uptime percentage into a useful service goal
Your website is reachable now. Was it reliable enough over the last month? A green current check cannot answer that question on its own. SLO monitoring gives availability measurements a target and a defined evaluation period, so you can review reliability with the evidence behind it.
MetriBeacon lets you create availability goals for HTTP monitors, inspect error budgets and generate reports for completed periods. This guide explains how to choose a first goal, understand uncertain results and make a short reliability review part of your routine.
What are an SLI, an SLO and an error budget?
An SLI, or service-level indicator, is the measurement used to describe a service. For this guide, it is the availability observed by your configured HTTP monitor. Its request and response conditions define what the check considers successful.
An SLO, or service-level objective, sets the target for that indicator over a period. An example is 99.9% availability over 30 days. This is an operational goal; creating it does not establish a contractual service guarantee.
The error budget expresses the amount of unsuccessful time compatible with that target. It helps you discuss how much reliability margin remains, instead of only reacting to the latest outage.
| Term | Question it answers | Example |
|---|---|---|
| Availability indicator | What does the configured check observe? | Whether an HTTP health request meets its conditions |
| Availability objective | What result are we aiming for? | 99.9% over 30 days |
| Error budget | How much unsuccessful time fits within the goal? | 43 minutes and 12 seconds for that full 30-day period |
| Data coverage | How much of the period has usable evidence? | Known time, with gaps shown separately |
All numbers here are illustrative. They are not measurements of MetriBeacon or a promise about your service.
Choose a monitor that represents an important outcome
Start with the website monitoring guide or API monitoring guide. Pick a request whose failure matters: a public application endpoint or a suitable health endpoint, for example.
Check what success actually means. A server that returns HTTP 200 with an error message may satisfy a status-only check. Use supported response conditions when they express the outcome you need. A goal based on the home page does not automatically measure checkout, sign-in or every API route.
Use the HTTP monitor's aggregate availability source offered by MetriBeacon. Do not add the durations from different checking locations together: several observations of the same minute do not create several minutes of service time. Keep a separate browser journey check when you also need evidence that an interaction works.
Create your first availability SLO in MetriBeacon
Follow the availability goals and reports setup guide. Owners and editors can manage goals and report schedules; viewers can read existing results.
- Select the workspace containing your HTTP monitor.
- Open Service level objectives and create a goal.
- Choose the monitor and a target percentage appropriate for the service.
- Select the evaluation period: seven, 30 or 90 days.
- Review availability and data coverage together before interpreting the result.
MetriBeacon starts with a 99.9% target and a 30-day period. Treat that as a starting configuration, not an automatically suitable commitment. Choose a target your team can explain and use when deciding what to improve.
For a newly created monitor, a long evaluation window can include time without observations. Let coverage build and inspect the uncertainty instead of assuming that the period before monitoring began was healthy.
Calculate what 99.9% means in practice
For a hypothetical complete 30-day period with no excluded maintenance, the arithmetic is:
30 days = 2,592,000 seconds
Allowed unsuccessful share = 100% − 99.9% = 0.1%
Error budget = 2,592,000 × 0.001 = 2,592 seconds
2,592 seconds = 43 minutes and 12 seconds
If that period contains exactly 20 minutes of confirmed downtime and no uncertain time, the remaining budget is 23 minutes and 12 seconds. This example uses the entire evaluation period; it is not a rule that each calendar month has the same number of minutes.
Use the displayed period and maintenance treatment for your actual result. Seven-day and 90-day goals have different time budgets, even at the same target percentage.
Read unknown time before declaring the goal met
Missing observations do not prove availability. MetriBeacon keeps successful, unsuccessful, unknown and unobserved time distinct, and expresses uncertainty through possible availability and budget ranges.
Extend the previous example with two hours of unknown or unobserved time. The confirmed 20-minute outage still fits within the 43-minute-and-12-second budget. However, the missing two hours could contain enough unsuccessful time to exceed it. You cannot confirm that the objective was met from those observations.
MetriBeacon confirms a goal only when the conservative availability result reaches the target. It confirms a breach when even the best possible result falls short. Other cases can remain uncertain. An old successful check does not fill a later gap; see the heartbeat and freshness article for that distinction.
When a result is uncertain, inspect the monitor's recent evidence and missing intervals before changing the goal. Raising a target or rounding the displayed percentage does not create the missing observations.
Decide how planned maintenance affects your review
Ordinary SLO evaluation includes maintenance. MetriBeacon also offers a separately identified maintenance-adjusted result that can exclude applicable maintenance intervals.
Choose the view that matches the question. The ordinary result describes the period including planned work. The adjusted result answers a different question with a different included duration. Label it clearly when sharing a number, and compare like periods using the same treatment.
A private maintenance schedule and a public explanation serve different purposes. Use an explicitly published status-page update when visitors need information about planned work or an incident.
Schedule daily, weekly or monthly availability reports
Open Reports → Schedules and choose a daily, weekly or monthly schedule. Check its local time, timezone and report language. Daily reports cover the previous day; weekly reports run on Monday for the previous week; monthly reports run on the first day for the previous month.
Generate the latest completed period once to inspect the result. Wait for the analysis to finish, then follow its completion link or open it from Report history. Repeatedly starting the same work does not make the analysis faster.
Completed reports preserve their period and source snapshot, including language and timezone. You can open them in your account and export CSV; they are retained for 90 days. Save an export deliberately if you need a longer record.
Email is optional. Only the workspace owner can enable it with a confirmed destination. Check the in-app report separately from its delivery: an email delivery problem does not by itself mean that report generation failed.
Make the report lead to one decision
Use a short recurring review: inspect the goal, coverage, confirmed downtime and incidents, then choose one improvement. A recurring deployment interruption might suggest a deployment change; a coverage gap might require fixing the monitor before judging the application.
Compare timestamps with the incident and delivery history guide and relevant changes. A shared timestamp is a starting point for investigation, not proof of the cause. Keep availability goals, immediate incident alarms and public communication as complementary tools.
Frequently asked questions about SLO monitoring
Does a 99.9% target guarantee 99.9% uptime?
No. It defines a goal for the configured monitor and period. It is not a contractual guarantee and does not cover unmeasured user journeys.
Does missing monitoring data count as successful time?
No. Unknown and unobserved time remain visible and can prevent a confirmed result, even if all recorded requests succeeded.
Can I use reports without email delivery?
Yes. Open completed reports in MetriBeacon and download CSV. Email is an optional owner-controlled delivery method.
Is SLO monitoring free in MetriBeacon?
MetriBeacon is free within its documented technical quotas. Check the usage and quota overview when planning monitors and recurring analysis.
Start with one service and one goal
Create a free MetriBeacon account, connect an HTTP monitor and define one availability objective. Review coverage, generate a completed-period report and choose one practical reliability improvement. The Website/API template below can start the HTTP monitor; add its SLO and report schedule separately through the app.
