Right-sizing will allow organisations to match the DCPI system to the load and reduce the energy consumption of DCPI equipment. For example, a typical system that is loaded at 30 percent of rating has an electrical cost per kW of IT load of about US$2,300 per kW per year. However, if the system were right-sized to the load, the electrical cost per kW of the IT load can fall to about US$1,440 per kW per year. This is a 38 percent savings in electrical costs alone. As future demand for server virtualisation and cloud computing projects increase, we definitely expect companies to right-size their data centres in an effort to prevent their infrastructure from being under-utilised.
Cloud Computing and Sustainability
For its new data centre, the National University of Singapore (NUS) has opted for the hybrid cooling approach, one of the few organisations in the region to do so.
Deployment of high density loads in IT environments can create areas of unpredictable cooling performance. The legacy approach to data centre cooling is inadequate for the medium and high density zones that the university has deployed in its new data centre facility, possibly giving rise to issues such as creation of hot spots, system downtime due to overheating and increased operational costs due to inefficient cooling.
A hybrid rack and row approach is used in this case - Schneider Electric's modular rack containment system was deployed to support the medium and high-density zones in the NUS data centre facility. Designed to contain clusters of high-density equipment up to 8kW and 15kW respectively, this standardised system acts as a self-contained mini data centre within a data centre with dedicated row-based cooling to prevent hot server exhaust air from escaping into the room effecting sensitive IT equipments.
Schneider Electric's InRow cooling is designed to closely couple the cooling with the IT heat load, allowing heat removal to be matched to the heat generation, improving efficiency and cooling predictability at both medium and high density zones. Heat is also removed at the source with this method, eliminating the mixing of hot and cold air streams which allows the system to predictably control IT inlet temperatures. In addition, the intelligent controls of InRow cooling actively adjust fan speed and chilled water flow to match the dynamic IT heat loads, enabling these units to conserve energy when servers are operating below design loads. As a result, Schneider Electric's InRow cooling can offer up to 30 percent increase in efficiency over traditional cooling architectures.
This deployment of the rack containment system, combined with InRow cooling, completely separates the supply and return air paths of IT equipment. This mixing of hot and cold air streams ensures predictable cooling behaviour and enables cooling capacity to be right-sized to the heat load. The modular nature of this setup also allows for quick deployment of high density clusters to address changing needs and future expansion with pay-as you grow architecture, providing NUS Computer Centre with the flexibility to scale and meet business requirements.
Sign up for CIO Asia eNewsletters.