Two circuits, one point of failure

Before comparing suppliers, name the events the business expects its connectivity to survive. A failed router, a cable damaged outside the building, a loss of mains power and an outage at an aggregation location are different problems. A design that addresses one may leave the others untouched.

Take a hypothetical warehouse with two circuits bought through separate service providers. If both fibres enter through the same duct and terminate beside the same power strip, the second invoice has not removed those dependencies. Either service might survive an individual carrier fault, while one excavation or local electrical fault could still stop both.

Draw the path from each application to the network. Include the LAN switch, router, power supply, carrier equipment, internal cable route and building entry. Continue through the access infrastructure as far as evidence allows. Mark each shared component and each unknown explicitly. A diagram with honest gaps is more useful than a reassuring line labelled backup.

Netavo provides connectivity across UK carriers and private WAN. Access to multiple carriers creates choices for a design, but a particular pair of circuits must still be assessed on its actual delivery. Carrier variety and physical separation are related questions, not interchangeable answers.

Begin with the premises. Will the routes use separate entries, risers and internal containment, or converge before reaching the communications room? Can a single damaged patch lead, local switch or flooded cabinet affect both? Where separation is required, make it part of the order discussion before work begins.

Then ask who operates each underlying access network. Different retail suppliers can use common infrastructure. Even different access operators may share ducts or another geographic constraint. Ask what is known about route separation, exchange or aggregation dependencies and the point at which the two services join a common network.

Request a written description of the proposed diversity and its limits. Evidence may be constrained by what an infrastructure owner can disclose, but uncertainty should remain visible. A sales statement that two routes are diverse needs a defined scope. It might refer only to equipment, only to a section of route or only to the carrier core.

For commercial property and facilities teams, landlord records and access permissions are part of this work. Separate external routes do little if the building requires both services to pass through one vulnerable cupboard. Agree who will preserve the intended separation when later building works move cables or equipment.

A surviving circuit is valuable only if the business can use it. Define how traffic moves after a failure and who owns that behaviour. Check whether the remaining path has enough capacity for essential applications, whether addressing changes affect access controls and whether existing sessions need to reconnect. These are design and acceptance questions, not assumptions about a product label.

Plan controlled tests around the failure scenarios. Disconnect one access path, then test loss of the relevant customer equipment separately. Observe a representative business transaction and incoming as well as outgoing communications. Record detection, recovery and application behaviour instead of treating a green router status as acceptance.

Power deserves its own test. Identify which devices depend on each supply and how long any backup arrangement can support the real load. The premises can have a healthy external connection while its router or switch is unavailable. Avoid claiming electrical independence simply because equipment has two power leads.

Keep a dated record of route evidence, shared risks, test results and accepted limitations. Revisit it after carrier changes, office moves or network redesigns. For a multi-site private WAN, examine central dependencies as well as branches, since a shared application location can become the common failure point. The objective is a clear statement of what the design survives and what operational response remains necessary.