ISP Diversity Audit: How to Prove Two Providers Don't Share the Same Fiber

You ordered two ISPs for redundancy and both invoices arrive on time, so the design feels complete. Then a contractor cuts the sidewalk outside your building and both circuits go dark at the same minute. The failure was not a coincidence. A large share of business fiber is leased, resold, and handed off through intermediaries, which means the two provider names on your contracts can easily resolve to the same physical cable, the same building entrance, and the same single point of failure. Paying twice for one path is the most expensive kind of redundancy to discover during an outage.

This guide gives you a repeatable audit to prove independence before you need it. You will verify diversity on paper, on foot, and on the wire, collect evidence that survives staff changes, and write contract language that keeps the diverse path diverse after the install crew leaves.

Why Two Logos Do Not Equal Two Networks

Carrier branding describes a billing relationship more than a physical one. Your building is reached by a local loop, which is the fiber from the street to your demarcation point, and that loop is frequently owned by a wholesaler or an incumbent local exchange carrier rather than the ISP whose name is on your contract. When that happens the ISP is reselling someone else’s fiber for the segment closest to you, and two different ISPs can be reselling the same wholesaler’s fiber without either sales team mentioning it. The result looks like diversity on a slide and behaves like a single circuit in the ground.

Wholesale leasing also collapses at the aggregation layer. Two carriers that do own their own metro fiber can still meet in the same central office, the same carrier hotel, or the same meet me room, and route your circuits through adjacent panels in the same facility. You have to test ownership and geography at every layer, not just the name on the contract, because sharing at any one of those layers defeats the purpose of buying a second provider.

Where Shared Fiber Actually Hides

Sharing hides in places a contract summary never shows. The five common points are the building entrance and riser, the conduit in the street, the local loop owner, the aggregation facility where access fiber meets metro fiber, and the building demarcation room where the handoff to your router lives. If both circuits use the same conduit for the last 200 meters, the same wall penetration, the same riser, or the same wholesaler to the central office, they share fate. The street cut, the building flood, or the wholesaler outage takes both down together.

The physical concentration point matters more than the carrier map. A network topology view that shows every circuit terminating at a site alongside its entry point and demarcation details makes a shared riser or a shared room visible the way a spreadsheet of circuit IDs never will. When you can see that both circuits arrive on the same floor through the same sleeve, you have found the real design problem, not a paperwork one.

Step One: The Paper Audit

Start with documents while you still have negotiating leverage. Ask each ISP for a letter that names the local loop owner for your address, states whether the building is on net or off net, and identifies the aggregation facility and the physical route from the street to the handoff. Request a route map at the level of the street and the building, not a metro wide marketing diagram, and treat any map marked illustrative as a placeholder until someone puts the specifics in writing. If the carrier cannot name the wholesaler, assume the wholesaler is shared until proven otherwise.

Cross check the paper with public data. Compare the carrier ASNs, look up the local exchange carrier records for the area, and ask the building manager for the riser and entrance records that show which providers have pulled fiber and through which penetrations. The riser manager often knows more about sharing than the ISP account team does, because the riser reflects what was actually installed rather than what was sold. Record every answer as a structured note on the circuit itself, so the evidence is queryable later rather than buried in an inbox that leaves when the owner of the inbox does.

Step Two: The Physical Audit

Paper tells you who claims to own the fiber. A walk tells you whether the claims are physically possible. Schedule a building walk with facilities and trace each circuit from the street to the demarcation point. Note the trench or duct on the street, the wall penetration, the conduit path, the riser, the room, the rack, and the power feed for each handoff. Photograph the labels at each point and measure whether the two paths share any sleeve, tray, or room. Two circuits that enter through opposite sides of the building, rise through different risers, and terminate in separate rooms on separate power have survived the physical portion of the audit.

Pay close attention to the demarcation room details. A diverse pair that lands on the same rack, the same patch panel row, or the same uninterruptible power supply shares a failure domain that has nothing to do with fiber. Record the rack, the U position, the patch panel, and the power feed for each termination, because the team that troubleshoots the outage will be looking for that mapping while the circuits are already down.

Step Three: The Logical Audit

Once the fiber is proven separate, prove the routed paths are separate as well. Run traceroutes from your edge to a stable set of destinations over each provider independently, and compare the second and third hops, the intermediate ASNs, and the geographic hints in reverse DNS. Two providers that are physically diverse but hand off to the same upstream ASN in the same facility will still correlate during a transit failure. Latency and jitter baselines for each path also help: circuits that ride the same fiber often show nearly identical latency to nearby destinations, while diverse fiber shows distinct latency profiles shaped by different distances to the aggregation point.

Optical evidence closes the loop where routing data is ambiguous. Ask for light levels and, where available, OTDR traces that show distance to splice points and confirm different physical distances to the central office. Matching OTDR signatures or identical fiber distances to the same exchange are strong hints that the underlying plant is shared, even when the provider names differ. The logical audit does not replace the physical one, it corroborates it, and the combination is what makes a diversity claim defensible when someone asks you to prove it rather than assert it.

Step Four: Contract Language and Acceptance Testing

Diverse fiber that is not protected in writing tends to become shared fiber at the first repair. Street works, building renovations, and emergency restorations give carriers a reason to temporarily rebuild your circuit on whatever plant is available, and temporary has a way of becoming permanent when nobody is checking. Your contract should state that the two circuits will not share the same conduit, entrance, local loop provider, or aggregation facility, and that any repair must restore the original diverse routing rather than leaving the circuits on the shared path that was convenient during the outage.

Tie that commitment to an acceptance test before the contract term starts. Define a window after installation where you exercise each circuit independently, verify failover, capture baseline traceroutes and latency over each path, and have both carriers sign that the diversity attestation matches what was built. If the diverse path costs extra, the test is what keeps you from paying for a promise that was fulfilled only on the day the sales engineer drew the diagram. Without a signed baseline, every future dispute about sharing becomes your word against a provider who has already been paid.

Documenting Diversity So It Survives Turnover

An audit that lives in email is an audit that expires when the team changes. Create a diversity record for the site that stores the evidence where the circuits are managed: the loop owner letter, the entrance and riser photos, the aggregation point names, the traceroute baselines, the optical readings, and the signed acceptance test. When that record sits alongside the circuit inventory and your network audit trail, with provider details, status, and site associations in one place, the next renewal does not require a new investigation, and the next outage does not require archaeology to answer whether the backup is actually independent.

That record also gives future changes a guardrail. When a circuit is replaced, upgraded, or rerouted, the diversity note makes the constraint visible before the work starts, so a well meaning change does not quietly collapse two independent paths into one. Teams that review circuit changes against the recorded entry points and loop owners catch violations during planning, when the fix is a different route, rather than during an outage, when the fix is an apology. For organizations that already track circuits as part of their broader documentation, keeping the diversity attestation linked to each circuit is the simplest way to make sure the audit you performed once keeps protecting the design every time the network changes.