Website uptime monitoring services periodically send requests to your domain, verifying if it responds with the expected status code. These checks aid in detecting and responding to outages and improving infrastructure over time. However, configuring checks to exclude reporting on regional outages can limit their effectiveness.

TL;DR: Complement your global test with a few regional tests in locations you care about to be informed about regional outages when they occur.

The usual setup

Let's assume you have a website that should be available globally.

The default setup on Status Cake is:

• Check Rate: 30s

• Confirmation Servers: 2

• Server location: random

This gives you global coverage and... good chance everything will be green even if you have regional outage happening for hours.

How StatusCake determines your website is down

Status Cake has 43 testing servers located around the globe (as of 02.2024). But only 3 servers in the South America. Let's assume your infrastructure provider has an 1-hour outage in South America.

For Status Cake to report that, also 2 consecutive checks from confirmation servers (which can be in different locations!) must report "down".

The probability of such event is:

$$\dfrac {2}{43} \times \dfrac {1}{42} \times \dfrac {1}{41}= \dfrac {1}{37023}$$

During a one-hour period, the service makes 120 calls (one every 30s). In practice, you'll never know about the outage affecting your customers in Brazil.

Set up complementary monitoring for regions you care about

The solution is simple:

• Set up uptime tests per each region you care about.

• Keep your global test (this will help you quickly see how serious the outage is and identify possible cause).

• Match regions with regions supported by your infrastructure provider.