Each recorded uptime observation is hashed together with the previous record's hash, forming a chain. Altering a past record breaks every hash after it, so a historical uptime percentage can be checked for tampering after the fact rather than taken on trust -- useful when uptime numbers back a contractual SLA.
Instead of a single poller, checks run from more than one network vantage point. A regional network issue or a single bad route then reads as a partial/vantage-specific signal instead of a false global-down alert.
Rather than firing a raw alert per failed check, related signals are summarized and prioritized, which cuts noise on fleets with many endpoints.