论文部分内容阅读
研究目的:数据中心服务器入风口温度升高将导致出风口温度随之升高,同时某些正在运作的服务器在高温环境下排放气体,可能会导致热点或硬件损坏。高出风口和热点造成制冷机制的负担。因此服务器可以被用于入风口温度灵敏度分析,且导致热点的位于高入风口区域的服务器可以被重新定位。创新要点:预测出风口温度作为入风口温度的参考。根据预测的出风口温度重新定位服务器,从而降低最大出风口温度并节省用于制冷的能量消耗。研究方法:根据能量守恒原则,提出服务器再定位算法(算法1),用于测试一组异构服务器。讨论不同测试组别下异构服务器再定位前后出入风口温度的时间响应以及CPU使用率(图2-28)。重要结论:所提热感知再定位方法应用于数据服务中心服务器可实现节能2.1 k W·h。再定位之后的服务器出风口温度同构化程度更高。对于每对再定位服务器,可以减少77%热点产生可能性。
Purpose of Study: Increasing the temperature of the air intake in the data center server will cause the air outlet temperature to increase, and some operating servers emit gas at high temperatures, which may result in hot spots or hardware damage. Higher than the outlet and hot spots caused by the burden of refrigeration mechanism. The server can therefore be used for air inlet temperature sensitivity analysis and the server in the hot air inlet area that caused the hot spot can be relocated. Innovative points: predict the outlet temperature as a reference to the inlet temperature. Reposition the server based on the predicted air outlet temperature, reducing the maximum air outlet temperature and saving energy for cooling. Research Methods: According to the principle of energy conservation, server relocation algorithm (Algorithm 1) is proposed to test a group of heterogeneous servers. Discusses the time response of inlet and outlet tuyeres before and after re-positioning of heterogeneous servers under different test groups and the CPU usage (Figure 2-28). Important conclusion: The proposed thermal relocation method applied to the data service center server achieves energy saving of 2.1 kW · h. Server relocation after the temperature of the outlet air temperature is higher degree of organization. For each pair of relocation servers, you reduce the chance of 77% of hotspots.