How uptime is measured

Each bar on this page is one day for one component. The percentage beside a component is its uptime over the last 90 days.

The daily figure

uptime = 1 - (major outage minutes + 0.3 x partial outage minutes) / 1440

A day has 1440 minutes. Today and the day a component was added count only the minutes so far. The 90-day figure adds up the same minutes over every day in the window.

What counts as down

• Major outage counts in full.
• Partial outage counts at 30%: some people or some requests were affected.
• Degraded performance counts as up: the service was slow but worked.
• Planned maintenance that we announced here in advance counts as up.

Bar colours

• No downtime
• Under 1 hour of downtime
• 1 hour or more of downtime
• No data
• How uptime is measured

Where the numbers come from

The figure comes from the status history shown on this page. A failing check changes a component's status only after it has failed from outside our network for at least 5 minutes, and a person reviews every incident. If we correct a status, the uptime figure is corrected with it.

This figure describes what we observed. It is not a service level agreement.