Reliability Methodology

Trust in recovery infrastructure requires clear definitions, repeatable measurement, and consistent disclosure. The Trust Center reports reliability and response indicators that reflect real platform behavior.

Driyu measures operational health from system checks and scan-flow telemetry. The same metrics used internally for reliability governance are exposed publicly in the Trust Center.

Operational means core systems are running within expected latency and availability thresholds.

Degraded means one or more services remain available but performance or reliability is reduced.

Outage means a core service is unavailable for normal operation. Status is based on recurring health checks and service-level signals.

A 99.99% annual availability objective corresponds to approximately 52 minutes and 34 seconds of downtime per year.

This value is a design objective, not a guarantee. Measured uptime is reported separately in historical records and may differ from the target.

STS measures elapsed time from scan initiation to owner-notification dispatch in completed recovery flows.

Lower STS indicates faster recovery routing. Driyu uses STS as a primary reliability indicator because recovery speed directly impacts real-world reunification outcomes.

Median STS represents typical performance. 95th percentile STS captures tail behavior during slower conditions.

Percentiles describe distribution. A 95th percentile STS means 95% of completed events are at or below that value.

Tail latency matters because recovery reliability is defined by worst-case responsiveness, not only average performance.

Global edge routing processes requests through geographically distributed regions to reduce network round-trip time.

Routing events to the nearest healthy region reduces latency variance and supports stable scan-to-notification performance.

Daily records summarize uptime and latency outcomes for historical visibility. The rolling 90-day view is generated from published reliability snapshots.

If historical coverage is still growing, the record may include partial days or null metric values while telemetry history is populated.

Public incident entries include material service events that could impact recovery performance or availability.

Each incident log includes impact and remediation context when available. Driyu maintains a public record as part of its transparency commitment.

Latency-triggered rollback means new infrastructure changes can be reverted when performance degrades beyond internal thresholds.

This protects recovery-critical flows by prioritizing stable routing performance during high-impact moments.