Data center design is complicated by rapid changes in technology and uncertainty about future capacity requirements....
Your question implies you are already thinking beyond initial system costs, and considering total cost of ownership, which is wise.
Energy costs and serviceability should be evaluated closely as they can have a significant impact on operating costs for years to come. Any data center cooling system being designed or modified today should balance underfloor cooling with supplemental cooling to optimize efficiency, reliability and flexibility. For example, a cooling system that balances the use of supplemental cooling units with underfloor cooling can achieve a 17 percent reduction in energy costs compared to using underfloor cooling alone.
But expansion costs are potentially even more significant. Equipment densities are continuing to rise and it is important to have a roadmap for dealing with higher densities to ensure you aren't limited in your ability to adopt new technology and make the most effective use of your data center space. Increasing rack spacing to accommodate higher densities is a costly strategy when facility costs are taken into account; significant savings can be achieved if the facility can scale to support higher density racks without reducing the number of racks being supported.
New technologies make it easier than ever to match cooling system capacity to current requirements, while easily adding capacity as needed. Newer floor mount precision cooling systems automatically adjust their capacity to room requirements without cycling compressors, increasing efficiency and reliability. This allows extra capacity to be designed into the system initially without a significant increase in energy costs.
Even more significant are the benefits of pre-piping the data center for high capacity cooling. Adding a high-capacity overhead cooling infrastructure makes it easy to add – or move – supplemental cooling modules as needed. Flexible tubing with quick-connect fittings allow cooling modules to be connected to the cooling infrastructure without the need for special tools or service support. This provides almost unlimited scalability and can extend the life of an existing data center and allow new facilities to support much higher densities than facilities constructed just five years ago.