Availability

99.5% a month, about 3.6 hours of downtime. This is a target, not a service-level agreement, and there are no service credits.

Each region is measured on its own API host. The figures are never combined.

Each month is added once it has ended.

Every 30 seconds, AWS Route 53 checks https://api.zeepler.com/v1/health and https://api.ca.zeepler.com/v1/health from three AWS regions: US East, US West and Europe. That request goes through the same API gateway as the label API, runs a function and reads from the same database, so it fails if any of them does.

The figure is the average share of Route 53's checkers that found the API healthy over the period. A checker marks it unhealthy after three failed checks in a row, so an interruption shorter than about 90 seconds may not show. Hours with no check data are counted neither as up nor as down, which is why the table shows how many hours were measured.

• It does not generate a label, so a fault in label generation alone would not show here.
• It measures whether the API answers, not how fast. We will publish a typical response time once we measure one.
• It checks from AWS, not from your network.
• It checks the two shared API hosts above. A host not listed here is not included.