How we calculate availability

This page explains where each figure in Availability, Status and Reports comes from. All times are Chile time.

What counts as downtime

  • Counts as downtime: the system does not respond or responds with an error, an operation fails, or a login flow cannot sign in.
  • Does not count as downtime: slowness (it is a warning), a certificate about to expire (notified separately) and anything during scheduled maintenance.
  • An incident opens after 2 failed checks in a row and closes on its own when the system responds correctly again. A single failure does not open an incident. This also applies to a brand-new monitor: one measurement alone does not alert.

Availability of a system

availability = time up / time measured × 100

time measured = length of the period − scheduled maintenance − measurement gaps
downtime      = union of the incidents of its active checks (excluding what falls within maintenance)
time up       = time measured − downtime
  • Measurement gaps: time when the platform could not measure. It is removed from the time measured; it is neither given away as available nor charged as downtime.
  • Union of incidents: if two checks of the same system fail at the same time, that time counts once.
  • Retired checks do not count. A retired system is measured only up to its retirement date.
  • Companies without an agreed target (signed up on the website): the time measured is the hours with checks, and downtime is the share of failed checks of their worst check over that time.

Overall availability

overall availability = Σ time up / Σ time measured × 100

It adds up every system that counts toward the target. It equals the average of the availabilities weighted by the time measured of each system: a system measured for half of the period weighs half. Systems pending confirmation are not counted.

Example for a 30-day month:

  • System A: measured 30 days, down 43 min → 99.90 %.
  • System B: measured 30 days, no downtime → 100 %.
  • System C: added mid-month (measured 15 days), down 2 h → 99.44 %.
  • Overall availability: 1 − (43 + 0 + 120 min) / (43,200 + 43,200 + 21,600 min) = 99.85 %. The simple average of the three would be 99.78 %; it is also shown for reference.

Target and downtime allowance

  • The target (for example 99.9 %) is the minimum availability agreed for each system.
  • The downtime allowance is what the target permits: time measured × (100 − target) / 100. At 99.9 % over 30 days it is 43 min 12 s.
  • Allowance used = downtime / downtime allowance. Above 100 % the target is missed.

Other figures

  • Checks OK: checks that responded correctly / checks performed. It shows intermittent failures that do not open an incident.
  • Mean time to recover (MTTR): average length of resolved incidents.
  • Mean time between failures (MTBF): time up / number of incidents.
  • Figures are shown with 2 decimals. The availability CSV has 3.

Sign in · Create account