Services and uptime targets (SLOs)

What a service is

A service is something people use. It is made of the checks it depends on. For example, "Shop website" might be made of:

• the internet line
• the firewall
• the web server's HTTPS port
• the website check

An uptime target (an SLO) says how available it should be over a period, for example 99.9% over 30 days. That leaves a downtime budget: about 43 minutes in 30 days at 99.9%. Uplivra shows how much of that budget is used and how fast it is going.

Add one

Reports › Services and uptime targets › Add a service.

1. Name it, and optionally describe it.
2. Set the target (for example 99.9), and the period: 7, 28, 30 or 90 days.
3. Choose when the service counts as up: - All of its checks are up. Use this for a chain, where every part is needed. - Any of its checks is up. Use this for spares, such as two domain controllers or a failover pair.
4. Tick the checks it depends on.
5. Leave alerts on to be warned when the downtime budget burns fast, runs low or runs out.

Reading the report

Each service shows:

• Now: up, down or degraded, from its checks' latest results.
• Availability over the period, measured each hour from its checks' results.
• Downtime budget left: the share of the period's allowed downtime that hasn't been used.
• Burn rate: how fast downtime is using the budget. 1× uses it up exactly at the end of the period.
• Daily availability: a bar per day, red when the day fell below the target.

Hours in approved planned maintenance are left out, as in the device availability report. Download everything as CSV for a monthly review.

Alerts

With alerts on, you get:

• A warning when the budget burns fast. That means 14× the sustainable rate over both the last hour and the last 6 hours: a real outage, not a blip.
• A warning when less than a quarter of the budget is left.
• An alert when the target is missed.

These alerts go to the same channels as your other alerts.