Testing in production is different than traditional preproduction testing. As demonstrated in the Creating alerts recipe, our production tests are implemented as monitors that are constantly testing the signals emitted by the system. However, if there is no traffic, then there is no signal, and so no tests to alert us to problems in any deployments that happen during these periods. This, in turn, decreases our confidence in these deployments. The solution is to generate steady synthetic traffic to fill in the gaps when there is no natural traffic.
Uptime checks are the simplest to put in place because they only make a single request. These should be included at a minimum, because they can be put in place quickly with little ...
