High-Density Computing: Efficient Versus Conventional Design
Data centers are expensive to build and operate. Large data centers cost $9–13/W to build [1] and can consume more than forty times, and up to over two hundred times, the amount of energy and resources consumed by a typical building [2], [3]. Therefore, space and energy considerations need to be accounted for when evaluating competing designs for high-performance computing (HPC) installations. This paper describes the results of an incremental cost and energy savings analysis conducted using data collected from a real-world case study to evaluate the impacts of efficient resource planning and implementing a total cost of ownership (TCO) model in the analysis of IT equipment and systems. The analysis presented demonstrates the advantages of using the latest technologies and IT strategies when planning the growth of new HPC installations at an enterprise level. The data also indicates an efficient design can significantly reduce the space, power, and cooling requirements of the HPC deployment while maintaining the performance and reliability criteria.