Most companies today are sensitized to the need for utmost reliability. They just need expert guidance on how to get and stay there. Unfortunately, smaller and mid-sized companies frequently turn to an IT software or hardware vendor not only for advice but to actually design data centers. Without in-house competencies, those vendors parcel the job out to several engineering sub-consultants who may be scattered across the country. Each of those axis points can introduce a separate concept of reliability, producing further loss of control and weakening overall reliability. In such cases, no single source is applying industry best practices to the facility as a whole, looking at all aspects of engineering, construction practices, facility maintenance, and operations. The requirements of the end-user should drive the design, but instead they usually get a cookie-cutter installation with overspending on one component at the expense of overall balance and reliability.
In one typical case, a data center was equipped with a very high-reliability UPS system (six-nines) surrounded by a three-nines electrical system with multiple single points of failure. In assessing and benchmarking their needs, we set a facility target of four-nines and reengineered and balanced the systems to follow suit. The “gold-plated” components of the UPS were deleted, but the overall reliability of the system was increased by balancing the capital costs across the electrical system.