RNP BUSINESS GUIDES / Keeping work moving

How does automatic Starlink failover work?

THE SHORT ANSWER

Automatic failover uses a router or firewall to detect an unusable primary internet connection and route traffic through Starlink. Both connections must be ready, and the router needs suitable health checks and traffic rules. Switching the route does not guarantee that existing calls, VPNs or payment sessions continue without interruption.

01 / Keeping work moving

What happens between failure and resumed work?

Separate the stages when measuring an outage
StageWhat happensWhat to record
Primary service becomes unusableA cable or provider problem affects trafficThe observed start of the fault
Health check detects failureThe router decides the WAN is unhealthyProbe behavior and detection delay
Traffic moves to backupNew traffic uses the second connectionSelected route and policy exceptions
Applications recoverSessions resume, reconnect or require staff actionTime until the essential task works
Primary service returnsThe router applies the return policyStability and any second interruption

02 / Keeping work moving

Why is a plugged-in cable not enough?

A WAN port can remain physically connected while the provider cannot carry useful internet traffic. The health check should detect the failure the business cares about. Ubiquiti’s UniFi documentation, for example, describes internet probes and configurable verification behavior; the exact controls depend on the gateway.

Ask the network administrator to document the chosen test targets, thresholds and return-to-primary behavior. An aggressive check can cause unnecessary switching, while a slow check prolongs the interruption. The correct choice depends on the service and the applications, so avoid copying unexplained settings from another site.

Source: Ubiquiti — WAN failover and load balancing.

03 / Keeping work moving

Will calls and VPN sessions stay connected?

Treat uninterrupted sessions as a test requirement, not an assumed feature. A route change can alter the public address or break a connection that an application must re-establish. Measure the user-visible result: did the call drop, did the VPN reconnect and did the transaction finish once?

The useful metric is time to usable work. A router may report that it switched quickly while a specialist application remains signed out. Include the application owner in commissioning so the handover explains any reconnect step and avoids promising seamless recovery that has not been demonstrated.

04 / Keeping work moving

Is failover the same as combining both connections?

No. Failover changes the path used when a connection is unhealthy. Load balancing distributes traffic according to the router’s policy; it does not automatically combine both services into one faster connection for every application. UniFi documents these as distinct operating modes.

Decide the business objective before selecting a mode. If the aim is an affordable emergency path, keeping bulk transfers off Starlink during normal operation may be appropriate. If the aim is distributing everyday work, data consumption and application routing need a different design.

Source: Ubiquiti — WAN failover and load balancing.

05 / Keeping work moving

What should the handover include?

  1. A diagram identifying primary WAN, Starlink WAN and shared power equipment.
  2. The approved list of traffic allowed during fallback.
  3. The observed application recovery time from a controlled test.
  4. A return-to-primary check and a procedure for unstable repeated switching.
  5. An owner for alerts, periodic retesting and changes to the router configuration.

About this guide

RNP’s planning recommendations are separate from vendor specifications. Sources were checked on October 15, 2026; current plan terms, availability and account-specific offers can change. Illustrative calculations are not installation measurements. RNP confirms project scope, price and local delivery availability after assessment.

PLAN THE INSTALLATION

Turn the question
into a clear brief.

Share the firewall model, primary circuit and applications that must recover.

Request a project review ↗

Scope, location, price and availability are reviewed before work is accepted.