基于温度预测的存储系统优化技术研究

发布时间：2018-06-27 07:49

本文选题：温度模型 + 温度预测　；参考：《华中科技大学》2012年硕士论文

【摘要】：对存储系统能耗的优化研究不仅是日益增长的数据量的客观需求，也是对绿色存储、节能减排号召的响应。在不考虑能耗的情况下单方面提高系统的性能，会导致电能的浪费，然而离开性能，单纯来降低系统能耗也显得意义甚微，因此，对存储系统的优化需要同时考虑性能与功耗两个因素，在节省能耗的基础上尽可能获得满足计算需求的性能。大量的实验数据表明，温度是影响存储系统性能的一个重要的影响因素，同时，温度也是能耗研究的一个不可忽视的条件，因此站在温度的角度来对存储系统的性能与功耗优化进行研究有一定的研究价值。针对存储系统温度过高会导致系统的性能下降，能耗上升这一问题，提出了一种温度预测的方法，，该方法可以准确的预测存储系统的温度变化从而有效的控制系统温度，防止温度过高。首先从存储系统的结构出发，设计并实现了一种系统温度与能耗数据的采集方法，有效采集存储系统的实时温度与能耗信息；其次，利用能量守恒定律结合热力学公式对系统的温度变化进行建模，同时采用递归最小二次方算法估算出系统温度模型中的参数，通过将采集到的温度历史数据带入模型，结合自适应调整对存储系统的温度变化进行预测；最后根据系统工作效率将系统的温度分为三个温度区域，针对这三个区域分别提出相应的系统温度控制方法，结合负载转移以及关盘操作来有效降低系统温度。对系统进行三个方面测试，首先对预测模型的准确性进行测试，结果显示温度预测模型可以准确预测出存储系统的温度变化趋势；其次对比分析了有温度控制和无温度控制时系统温度的变化曲线，发现温度控制可以有效的将系统温度控制在最优温度范围内；最后对系统功耗优化进行测试，测试表明温度控制对系统功耗的降低效果符合预期。
[Abstract]:The research on energy consumption optimization of storage system is not only the objective demand of increasing data volume, but also the response to the call of green storage, energy saving and emission reduction. Without considering the energy consumption, the unilateral improvement of the system performance will lead to the waste of electric energy. However, it is of little significance to reduce the system energy consumption simply by leaving the performance. Both performance and power consumption should be taken into account in the optimization of storage system. On the basis of saving energy consumption, the performance of computing requirements should be obtained as much as possible. A large number of experimental data show that temperature is an important factor affecting the performance of storage system, and temperature is also a condition that can not be ignored in the study of energy consumption. Therefore, it is valuable to study the performance and power optimization of storage system from the point of view of temperature. In view of the problem that excessive temperature of storage system will lead to the deterioration of system performance and increase of energy consumption, a temperature prediction method is proposed, which can accurately predict the temperature change of storage system and effectively control system temperature. Prevent excessive temperature. First of all, from the structure of the storage system, a data acquisition method of system temperature and energy consumption is designed and implemented, which effectively collects the real-time temperature and energy consumption information of the storage system. The temperature change of the system is modeled by energy conservation law combined with thermodynamic formula. At the same time, the parameters of the system temperature model are estimated by using the recursive least square algorithm, and the collected temperature historical data are brought into the model. Finally, the temperature of the system is divided into three temperature regions according to the working efficiency of the system, and the corresponding system temperature control method is put forward for the three regions. Combined with load transfer and diskette operation to effectively reduce system temperature. The system is tested in three aspects. Firstly, the accuracy of the prediction model is tested. The results show that the temperature prediction model can accurately predict the temperature trend of the storage system. Secondly, the temperature variation curves of the system with and without temperature control are compared and analyzed. It is found that temperature control can effectively control the system temperature within the optimal temperature range. Finally, the power consumption optimization of the system is tested. The test results show that the temperature control can reduce the power consumption of the system.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP333

【参考文献】