“Two carriers” is often used as shorthand for resilience. It is a useful starting point, but it is not proof. Services with different logos can share the same physical or logical failure domain for part of the journey.
The practical definition
Routes are diverse only to the extent that one credible incident cannot reasonably remove both services.
The answer depends on the failure being considered: a cable strike, chamber fire, bridge closure, power loss, node failure, wholesale fault or operational error.
1. Define the failure domains that matter
Absolute independence is rarely achievable or economically rational. The aim is to identify shared dependencies, understand their consequence and decide which must be removed. A campus serving mission-critical workloads may require a different standard from a secondary office connection.
Review diversity across at least six layers:
- External physical route: ducts, poles, chambers, cables, joints, bridges, tunnels, railway crossings and river crossings.
- Site access: boundary chambers, duct banks, building entries, risers and meet-me rooms.
- Network nodes: local exchanges, aggregation POPs, optical line systems, routers and handover sites.
- Underlying supplier: wholesale tails, leased fibre, PIA or other third-party infrastructure used by the retail carrier.
- Power and environment: shared power feeds, cooling, racks or building systems at a handover point.
- Operations: common maintenance teams, change windows, management platforms or escalation paths.
2. Look for sharing where routes naturally converge
Even well-designed routes often approach the same constrained locations. Those locations deserve disproportionate attention because one event can affect a wide area.
- A single bridge, culvert or rail crossing on the only practical corridor.
- A shared chamber outside the site before paths split.
- Two carriers using the same Openreach or third-party duct section.
- Separate access cables returning to the same local exchange or metro node.
- Distinct wavelengths carried over the same underlying fibre pair.
- Two services delivered through the same carrier hotel, rack, optical platform or power distribution.
- Both routes following the same road for a material distance, even in different ducts.
Some sharing may be acceptable. The key is that it is known, assessed and not obscured by a high-level “diverse” label.
3. Use an evidence hierarchy
Providers differ in how much route information they will disclose. Security and commercial restrictions are legitimate, but a critical customer still needs enough assurance to make a decision. Record the strength of each piece of evidence.
4. Ask precise questions rather than “is it diverse?”
A binary question invites a binary answer. More useful questions expose the boundaries of the claim:
- At which physical point do the two routes become separate?
- Do they share any duct, chamber, pole line, cable, bridge, tunnel or specialist crossing?
- Does either carrier use another operator for the local access or backbone segment?
- Which aggregation and handover nodes are used?
- Are the services on separate optical systems, line cards, routers, racks and power feeds?
- Can the route change after delivery, and how will the customer be notified?
- What evidence will be available at design approval and at service acceptance?
- What is the remedy if the delivered service does not meet the accepted route design?
5. Protect diversity at the site boundary
External diversity is commonly lost in the final metres. If both carriers use the same entrance chamber and duct into one meet-me room, a local excavation, chamber fire or internal incident can defeat the investment made in separate external routes.
A robust site strategy can include:
- Geographically separated boundary entry points.
- Separate chambers and duct banks to the buildings.
- Independent wall penetrations and internal pathways.
- Separate meet-me rooms or clearly separated carrier areas where justified.
- Documented cross-connect and internal fibre paths.
- Independent power and environmental support for critical carrier equipment.
6. Produce a shared-dependency map
A useful diversity review shows where the paths are independent, where evidence is incomplete and where they knowingly share infrastructure. This is often more decision-useful than a visually impressive map with unexplained coloured lines.
Illustrative dependency register
7. Choose the right diversity pattern
Several patterns can be valid depending on risk and budget:
- Dual carrier, dual route: separate providers and physical paths to separate handover nodes.
- Single carrier, contractually diverse routes: one provider manages two separately engineered services, simplifying operations but concentrating supplier risk.
- Owned dark fibre plus managed service: different operating models reduce common technical and commercial dependencies.
- Neutral access route to multiple carriers: the site controls or accesses a shared route to a carrier-rich location, then selects independent onward networks.
- Geographically separated campuses or gateways: resilience is created at a wider architectural level rather than only at one building.
The architecture should distinguish route resilience, carrier resilience, equipment resilience and destination resilience. One does not automatically provide the others.
8. Manage diversity throughout the service life
Networks change. A carrier may reroute a service after a fault, optimisation or acquisition. Construction elsewhere may place originally separate routes into a shared duct. The diversity position therefore needs governance after acceptance.
- Keep route and dependency records under configuration control.
- Require notice of material route or underlying-supplier changes where commercially achievable.
- Revalidate following major repairs, migrations, contract renewals or network consolidation.
- Include diversity in operational reviews and incident post-mortems.
- Test failover and routing behaviour, not only optical continuity.
9. State residual risk honestly
A route-diversity report should not imply certainty that the evidence cannot support. It should identify confirmed separation, known shared points, provider assertions, unverified sections and the consequence of each remaining dependency.
That allows the project to make a rational choice: accept the residual risk, seek stronger contractual assurance, undertake a physical survey, change carrier, move the interconnection point or construct a missing neutral segment.
The decision rule