Separate the failure scenarios
An internet circuit fault, a carrier fault, a local switch failure and a loss of mains power are different events. A backup internet link will not repair a failed switch. A cloud server still running does not prove that a room phone can reach it.
Draw the path for room-to-reception calls, incoming reservations and external staff calls. Include handsets, gateways, switches, router, PBX, carrier and power supplies. Mark which components are shared so a single failure is visible.
Check what the backup really covers
A UPS needs to support the relevant devices, not only the PBX. Record its load and test runtime with the responsible facilities team. Include equipment in remote cabinets and any network devices needed to reach the telephone service.
A second internet service needs a tested changeover path. Ask whether the providers share physical infrastructure, whether the backup can carry the expected load and whether the carrier accepts calls from the backup connection. Existing calls may be interrupted when a route changes.
Agree a usable fallback for staff
A provider-side diversion to an agreed destination may help incoming calls, where the carrier supports it. Confirm how it is activated, who is authorised and how callers are handled when that destination is busy.
Keep an independently usable contact method for the duty team where needed, with clear instructions and maintained equipment. Emergency calling, location information and public-number behaviour must be verified with the relevant provider for the actual configuration.
- Scenario: name the component or service that fails.
- Expected behaviour: state which call types remain available.
- Staff action: name the person, procedure and alternative contact route.
- Test record: record time, observed result, limitations and recovery.
Rehearse and revisit
Arrange controlled tests with the hotel and affected providers. Start with an agreed maintenance window, a recovery method and a way to communicate independently of the system under test.
Retest after relevant changes to routers, carriers, power equipment or PBX configuration. The result is evidence for a specific scenario on a specific date, not an uptime guarantee.