Synthetic monitoring is a monitoring method that periodically sends scripted requests imitating real user access from the outside and continuously measures a service's availability and response performance. Because problems can be detected without waiting for real users' actions, it enables early discovery of outages and a proactive response.
In the mechanism of synthetic monitoring, monitoring nodes placed around the world send HTTP requests to the target service and verify the response time, the status code, and the contents of the response body. Scenarios can be defined ranging from simple up/down checks (ping) to complex ones such as following login flows and redirect chains. Examples of tools include Datadog Synthetic Monitoring, Amazon CloudWatch Synthetics, and Pingdom.
For URL shorteners, there are designs that continuously measure the P95 latency of redirects and raise an alert when it exceeds a threshold (e.g., 200ms). Monitoring from multiple regions also makes it possible to detect DNS resolution delays and CDN outages that occur only in specific areas.
As an implementation consideration, a mechanism is needed that identifies monitoring requests by User-Agent or headers and excludes them from aggregation so that they do not mix into actual access statistics. The monitoring interval varies depending on the service and the tool.