Home › Guides

Planning a data centre for a telecom operator

By G2MAX · Updated 1 November 2026

A telecom operator's data centre must connect the power plan to the service plan. Network functions, customer hosting and internal applications can have different availability, latency and security needs even when they share a building. Begin with a service inventory and the paths those services depend on. A resilient room is of limited use if every service still relies on one external cable route.

Define what will run at the site

List the intended services, their initial demand and their growth assumptions. Separate network equipment, general computing, storage and any customer racks. Record power type, voltage, airflow, rack dimensions and expected heat output for each group. Some telecom equipment uses DC supplies while computing equipment uses AC; do not assume a single distribution arrangement fits everything.

Translate business requirements into measurable operating limits. Define tolerated interruption, restoration priorities and maintenance windows. Identify services that can move to another location and those that cannot. A duplicated application may tolerate a local equipment event differently from a service tied to one access network. These distinctions help allocate resilience where it supports the actual service.

Trace routes beyond the building

Document fibre entry points, ducts, chambers and the upstream paths serving the site. Two carriers or two cable labels do not prove physical diversity. Ask where routes share trenches, bridges, exchanges or other infrastructure. Record evidence and unresolved questions instead of treating commercial separation as proof that the routes cannot fail together.

Inside the building, plan the meet-me area, cross-connect routes, patching space and access rules. Keep A and B paths separate where the service design depends on that separation. Leave space for changes without forcing technicians to work across live connections. Agree labelling and records before installation; reliable fault finding depends on knowing which service each connection supports.

Start power design from measured demand

Use measured loads where existing equipment is being moved, with an allowance for growth that the owner can explain. Nameplate totals can differ substantially from expected demand, but diversity assumptions need evidence. Make a rack schedule that separates opening load, contracted capacity and ultimate capacity. That avoids selling or allocating capacity that the power and cooling systems cannot actually deliver.

For early planning, convert the available facility power into an IT allowance after cooling and losses. Then distribute that allowance across service groups and phases. Check whether DC plant, batteries and AC UPS systems introduce separate room requirements or maintenance regimes. Shared services such as controls, fire protection and cooling should appear in the same interface register as the rack supplies.

Design for remote operation and local response

Many operator sites have limited staff on site. Define which alarms need immediate action, which can wait and who is permitted to switch equipment. Ensure that critical alarms still reach operators when the main network path fails. Remote access needs an agreed security and recovery process rather than a permanently open maintenance connection.

Provide local indications and safe manual procedures for loss of remote control. Keep spare parts and trained response resources appropriate to travel times and the equipment used. A remote monitoring screen cannot replace someone able to inspect a leak, operate an isolation point or confirm that a generator has accepted load. Test escalation with the people who will actually receive the call.

Plan migration as part of the project

A new room does not by itself create a safe migration. Inventory dependencies, addressing, cross-connects and rollback requirements. Group moves into controlled batches with clear acceptance checks. Agree which existing systems must remain active until a service has been tested in its new location and the owner has accepted the result.

During integrated testing, involve both facility and network teams. Prove that power transfers and cooling events do not produce unexpected service failures. Check time synchronisation and event records so teams can reconstruct a fault in order. Close the project with as-built routes, circuit schedules, monitoring ownership and an agreed process for future changes.

Keep growth within operational limits

Reserve useful expansion space at the rack, distribution and network layers. A spare rack position is not usable growth capacity if busways, cooling or patch fields are full. Review capacity using the tightest constraint and update that record when customers or network functions are added. Link each new load to an approved capacity allocation and a documented installation path.

Questions