The conclusion is direct: a colocation facility sells uptime, and its largest threat is power, so the right backbone is a modular UPS that scales in 25 to 50 kVA building blocks, holds efficiency above 96 percent at partial load, and delivers N+1 redundancy without the cost of a fully duplicated train. That combination is what lets a provider hold a 99.999 percent availability target while filling racks gradually.
Tenants pay for guaranteed availability measured in nines; a 99.999 percent target allows only about five minutes of downtime per year. A single voltage sag, a failed rectifier, or a maintenance event on a monolithic UPS can breach that budget and trigger contractual penalties. Worse, a traditional single large UPS either runs near-empty, wasting energy, or forces the provider to overbuild capacity years before tenants arrive and pay for it.
Modular UPS frames accept hot-swappable power modules, typically 25, 50 or 100 kVA each, inside a shared cabinet. As a hall grows from 100 kVA to 800 kVA, the provider adds modules rather than installing a second system. The capital tracks the actual load instead of a five-year forecast, which protects return on assets and shortens the time to revenue on every cabinet brought online.
Because each module is independent, N+1 is achieved by installing one spare module rather than a parallel full-size unit. A frame sized for 400 kVA can carry 450 kVA of modules, so the loss of one simply sheds the redundant capacity. Tenants receive the resilience they specify, and the provider avoids the floor space, cooling and expense of a 2N build for every hall.
Colocation halls rarely run at full design load, especially early in their life. Modular topology holds 96 to 97 percent efficiency in double-conversion mode across a wide load band and stays above 95 percent even at 30 percent load. On a 1 MW critical load, a three-point efficiency gain versus an older system is roughly 30 kW of saved loss, cutting both the power bill and the heat the cooling plant must remove.
When a module needs attention, the technician slides it out while the remaining modules carry the load. There is no scheduled downtime window and no transfer to raw mains. For a colocation operator the promise that maintenance never touches a tenant rack is itself a sellable feature that supports the SLA.
The practical play is to size the frame for the hall's eventual peak, populate it with modules for today's load plus one spare, and grow block by block. Our reference configurations for data-centre and colocation deployments are outlined at https://www.upsboss.com/data-center/, and the engineering notes at https://www.upsboss.com/faq/ answer the common questions operators raise during procurement.
Key takeaway: colocation power is won on scalability, partial-load efficiency and online serviceability. A modular UPS delivers N+1 resilience and 96 percent-plus efficiency in 25 to 100 kVA steps, so the provider pays for capacity only as tenants arrive.
Does modular UPS really reach 99.999 percent availability?
The UPS alone does not guarantee five nines, but hot-swap modules and N+1 design remove planned maintenance and single-module failure as causes of downtime, which is the largest share of avoidable outages.
Is modular more expensive than a single large unit?
Per kVA it can cost more upfront, but because capacity tracks real demand and avoids overbuild, the total cost of owned-and-used capacity is usually lower across the hall's life.
Can modular UPS share a battery with the IT load?
In most designs each module has its own battery string or draws from a shared centralized bank; either way the configuration should be confirmed against the fire and ventilation rules for the room.
Contact: Frank Zhang
Phone: +86-135 5688 8641
Email: frank@upsboss.com
Add: Jufeng Road, Guangming Street, Guangming District, Shenzhen City, Guangdong Province, China
