← Back to the chapter · Previous detail

Technical detail · Sigfox Continuity and Service Risks

Sigfox continuity register

18.10 Sigfox Continuity Register

The continuity register is a living release artifact. It converts managed-service dependency into testable evidence, owners, monitoring, incident classes, and exit triggers.

Start the register before the pilot expands. Each row should say which part of the path is being trusted, what evidence proves it today, who owns the response, and what event reopens the decision. If any column is blank, the deployment may still be useful as a trial, but it is not yet a clean scale decision.

Risk fieldEvidence to keepOwner to nameRecheck trigger
Service receptionRepresentative messages from installed devices with timestamp, signal context, device identity, and missing-message checks.Field operations and service-monitoring owner.Coverage change, deployment region change, antenna/enclosure change, or missed-message threshold.
Callback pathDecoder output, callback status, backend response, storage freshness, retry behavior, and alert creation.Platform/backend owner and application owner.Endpoint change, authentication expiry, decoder change, payload version change, or backend incident.
Incident triageRunbook that separates device fault, service gap, callback failure, application rejection, and unowned action.Support lead and release owner.Repeated unknown failures, delayed triage, unclear owner, or customer-visible data gap.
Exit readinessPayload contract, device identifiers, data export path, replacement criteria, and fallback network options.Product owner and architecture owner.Service withdrawal, commercial change, region loss, support degradation, or unacceptable SLA gap.

Watch what users need. Alert on application freshness, missing messages, callback failures, and stale devices instead of only service-dashboard counts.

Classify the failed layer. Separate radio/service reception, decoder, callback, backend, application, support, and ownership failures during incident review.

Reopen old decisions. Require a review when cadence, firmware, payload format, device region, operator terms, callback endpoint, or product promise changes.

Use thresholds that match the promise instead of generic uptime language. A waste-bin sensor may tolerate missed reports for hours; a flood-warning sensor may need much tighter freshness and escalation. The register should make that difference visible so a future reviewer does not inherit an approval that was only valid for a weaker use case.

The register is also the incident handoff. During a failure, the support team should be able to point at the last proven evidence point, assign the right owner, and decide whether the release record stays valid, needs a limit, or must be reopened.

← Back to the chapter · Previous detail