The data centre fibre connectivity checklist

Power, planning and land usually dominate early data centre site selection. Fibre can appear simpler because a carrier map shows coverage nearby. In practice, the difference between “network in the area” and a deliverable, diverse high-capacity solution can be substantial.

1. Define the connectivity requirement before asking for quotes

Carrier responses will only be comparable if the project team has defined the outcome. Start with the day-one requirement and the expected expansion path rather than requesting a generic “fibre quote”.

• Required day-one and ultimate capacity, for example diverse 100 Gbps with a credible path to 400 Gbps or more.
• Number of physically separate external routes and site entrances.
• Whether the site needs dark fibre, managed wavelengths, Ethernet, IP transit or a combination.
• Target destinations: London, Manchester, Glasgow, Newcastle, cloud on-ramps, internet exchanges, other campuses or customer locations.
• Operational model, including who owns the optical equipment and who manages faults.
• Target ready-for-service date and any phased energisation programme.

A requirement that is too vague produces attractive but incomparable proposals. One carrier may price a protected managed service, another an unprotected tail, and another a construction contribution that excludes the final site entry.

2. Establish what “nearby network” actually means

A cable passing close to the land boundary does not necessarily create an available service. The network may be a long-distance route with no practical breakout point, may lack spare fibre, may be owned by a wholesaler rather than the responding carrier, or may sit on the wrong side of a difficult crossing.

For each credible network, establish:

• The nearest useful point of presence, joint, chamber or interconnection opportunity.
• Whether the responding provider owns the route or relies on third-party infrastructure.
• The likely distance and construction method from the network to the proposed meet-me point.
• Any motorway, railway, river, bridge, protected land or private-land constraint between the two.
• Whether capacity is available now and whether it can be reserved for the programme.

3. Test capacity, product and upgrade path

Dark fibre, wavelengths and Ethernet services solve different problems. The lowest monthly charge may transfer more operating responsibility to the data centre; the highest charge may include protection or equipment the project does not need.

Also test how the solution scales. A service that meets phase one but requires a complete route replacement for phase two is not necessarily the lowest-cost option.

4. Prove physical route diversity

Two suppliers can still fail together. Shared risk commonly appears in the access duct, a bridge or railway crossing, a local aggregation node, a wholesale tail or a single building entry. Ask for evidence at a level proportionate to the criticality of the site.

• Separate route drawings or clearly stated route descriptions.
• Identification of shared third-party duct or fibre.
• Separate site entrances, chambers and internal pathways.
• Named aggregation or handover nodes where disclosure is possible.
• Contractual treatment if the delivered route differs from the accepted design.

The objective is not perfect cartographic certainty at the earliest stage. It is to identify what is known, what is asserted by a provider, what remains unverified and what must be surveyed before investment or acceptance.

5. Design the external and internal site entry together

A diverse external network can be undermined by both services entering the same chamber, duct bank, wall penetration, meet-me room or powered distribution area. Connectivity therefore needs an interface with the architectural, civil, M&E, security and operational design.

• Primary and secondary site boundaries and carrier access routes.
• Duct capacity, draw pits, chambers and future spare paths.
• Separate building penetrations and fire-stopping strategy.
• Meet-me-room location, access control and carrier demarcation.
• Internal fibre paths to network rooms and customer areas.
• Space, power, cooling and security for carrier equipment.

6. Quantify construction, land and approval risk

The last kilometre often determines the real delivery date. A credible programme should distinguish desktop assumptions from surveyed route, public-highway work, private-land rights, third-party approvals, traffic management and specialist crossings.

The cost plan should state what is included, the basis of route length and reinstatement, whether traffic management and statutory fees are allowed for, and what contingency remains for unknown utilities or authority requirements.

7. Treat latency as a route question, not a postcode calculation

Geographic distance provides only a lower-bound indication. Actual latency depends on where the carrier hands over, how the route is engineered, whether traffic travels through an indirect core node and where the cloud or internet service is interconnected.

For material destinations, request an indicative route and measured performance from a representative node where possible. Record whether the figure is round-trip latency, an estimate or an existing-service measurement.

8. Normalise commercial proposals

Compare the whole offer, not only monthly recurring charge. The decision model should include:

• Non-recurring construction and connection charges.
• Monthly or annual recurring charges and indexation.
• Minimum term, notice and renewal mechanism.
• Excess-construction risk and survey conditions.
• Service credits, repair targets and escalation.
• Capacity upgrade pricing and lead time.
• Wayleave, access and equipment responsibilities.
• Cost of optical equipment, support and spares where applicable.

9. Design for the fault that will eventually occur

Operational readiness is often left until handover. The project should know who receives alarms, who can access chambers and meet-me rooms, how an external cable failure is escalated, what spares are held and how route records will be maintained.

Acceptance evidence should include optical test results, as-built drawings, fibre schedules, demarcation records, contacts, escalation paths and confirmation that the delivered route matches the agreed resilience design.

10. Convert the evidence into a clear site decision

A useful due diligence report does not merely list carriers. It should state the recommended route strategy, confidence level, key assumptions, capital and recurring cost range, delivery programme, material risks and the next action required to close each uncertainty.

Start broad, then spend where the evidence matters

At the land-search stage, a high-level desktop screen may be sufficient to remove weak locations. Detailed provider engagement, route surveys and design should then be concentrated on the shortlist. That staged approach protects both speed and diligence.