A fleet of agents across regions and pools probes every monitor in parallel, so a regional outage looks like a regional outage. Run the cloud probes, drop your own agents inside a VPC, or both.
Define your own regions and pools in the admin console; agents self-register with a single-use token and heartbeat from there, whether they're our cloud probes or your own workers running inside a VPC to reach an internal endpoint. Every monitor is checked from every region you choose in parallel, so a regional outage reads as regional rather than as a mystery blip — and ICMP, TCP and HTTP checks run side by side from each node.