One of the issues with the traditional ways of designing and building data centers is that very often the performance of the data center infrastructure does not become apparent until the facility is complete and load tested, at which point an indication of the overall capacity of the data hall can be assessed. Since at the very beginning of a new data center’s lifecycle its performance can be as uncertain as the future requirements of IT, a worst case final load requirement is dialled into the design and build. Because the cost and operational penalties for running out of capacity during the data center’s life are steep, significant CapEx and OpEx are incurred by overbuilding.
In addition, actual performance can be lost and further cost accrued because of the difficulties of integrating different data center sub-systems and components with one another. For example, the individual parts that comprise chiller plant may achieve published Coefficient of Performance (COP) standards, however, most chiller plants achieve a much lower COP in operation because of problems encountered when attempting to integrate the controls of its various parts. The ineffectiveness of custom designed controls implemented in the field often means significantly less time operating in economizer mode with consequent increase in energy cost and impact upon PUE.
At DatacenterDynamics London 2013, Schneider Electric displayed I caught up with Scott Neal, product marketing director for prefabricated solutions to discuss how the approach can help resolve the issue of predictability in data center performance. Scott said, “If we look at the way traditional data centers are designed today and get into the long life cycle of the typical data center design, build and implementation, the products themselves were essentially designed by an engineering firm or a consultant, installed by a contractor and then there’s several sub-contractors, so there’s lots of hands involved.”
“That process can still exist except for the value of us providing a prefabricated module is that we’ve engineered the module, we’ve engineered the equipment side of the module to work together, we’ve ultimately created a whole solution from a reference design. So ultimately what the customer gets in the end is essentially what it was designed for in the beginning, and there’s a much, much higher chance of that working out in their favor, versus a long life cycle, more hands touching in and ultimately things changing over time and not achieving the level of performance that you thought you were going to get a year ago.”
So, in the first instance, pre-assembly, integration and testing of prefabricated modules means that performance characteristics of the modules can be published and utilised confidently. In addition, the issues associated with field implementation and integration are overcome before the modules leave the factory, lowering the cost of on-site services and time of deployment. In addition, a published library of Data Center Reference Designs for prefabricated data center modules allows the consumer to specify and the manufacturer to publish expected efficiencies based on real measurements of the design. Predictable efficiency is especially attractive for businesses which have a focus on energy efficiency initiatives.
More details regarding cost and performance of prefabricated data center modules can be found in white paper #164, “TCO Analysis of a Traditional Data Center vs. a Scalable, Containerized Data Center”, available for free download from the Schneider Electric website.