by : Alison Campbell

On the one hand, there's the management issue of controlling several smaller units attached to vital equipment or each separate network. On the other, the cost and reliability implications of having to install a dedicated ring main system for power management; if the central UPS fails, so does the whole system!

When planning business continuity, a centralised UPS is normally more powerful, functional, sophisticated and automated than a solution based on a distributed architecture. What's more, it's usually housed away from the main data centre. Distributed UPS, however, is simpler and less costly to install; vital equipment can be protected independently, and thus it is shielded from a central power outage.

Deciding on a

When planning power and business continuity, one of the main issues is to protect against 'single-points-of-failure'. In practice, both centralized and distributed UPS solutions work well - providing the system is designed with appropriate redundancy and a built-in maintenance by-pass (allowing UPS maintenance without loss of power to the load).

A centralised uninterruptible power supply usually forms part of the facility, as it has to be housed securely away from the main business. In the past, these systems were only meant to protect against short-term power outage but this is now changing. With modern UPS regarded as a lifeline service, battery back-up, failover, redundancy and maintenance by-pass are all now an integral part of their design. Centralised solutions use a single UPS to feed a custom built ring main, which supports numerous items of critical load or dedicated networks.

The alternative is a distributed/decentralised system, with each piece of data centre equipment powered by a smaller individual UPS.

Load utilisation and your uninterruptible power supply

In terms of load utilisation and battery capacity, a centralised uninterruptible power supply can be more versatile and cost efficient. However, the installation of a dedicated ring main can be expensive - especially if the loads are spread around a building. While distributed solutions may not optimise load capacity and extended run time, the effects of UPS failure may be reduced when compared to the business continuity implications of a poorly-designed, failed centralised UPS.

Distributed UPS means that the business is grouped into data processing clusters, each with dedicated UPS protection, giving several smaller power plants spread across the whole operation. Distributed UPS often require lower capital outlay and installation costs. Some suppliers claim higher overall operating efficiencies because the number of conversions from mains AC to DC is reduced - leading to energy savings.

The primary advantage of centralized UPS

Smaller UPS generate less heat than larger ones, leading to greater efficiency. Distributed systems also reduce the total number of power conversion steps; consequently, electronic components and circuit complexity are both reduced. Distributed solutions, however, generally have more components, which could increase the likelihood of breakdown. As each data centre processing cluster has its own uninterruptible power supply, confusion about which equipment is protected against power outage by which UPS is eliminated. Likewise, maintenance and battery changes are localised to individual clusters thus reducing downtime elsewhere in the business.

The primary advantage of centralised UPS over smaller, distributed UPS lies within batteries. Smaller UPS contain sealed lead-acid batteries - with a typical design life of five years or less. The cost implications of this depend on the number of UPS clusters. Larger batteries, used in centralised UPS, have longer design lives.

Another business continuity benefit of centralised UPS is that, being housed away from 'busy' areas of the building, it is less easily disrupted, accidentally damaged or maliciously interfered with.

Always invest in expert UPS consultancy

In conclusion, distributed UPS, though easier and less costly to expand, can cost more per-kilowatt than a centralised system. Centralised systems, however, are more expensive initially to build and must be carefully designed to meet future business continuity needs. Large-scale Power-over-Ethernet (PoE) deployments of IP phones, wireless access points, internet security cameras and other peripherals may favour a centralised uninterruptible power supply solution through lower whole-life costs.

When planning power and business continuity, the wisest option is to treat each network on a case-by-case basis and always invest in expert UPS and business continuity consultancy.