温度属性对存储设备可靠性影响的研究
发布时间：2018-07-28 18:21
【摘要】：对数据中心服务器的损坏情况进行跟踪统计发现：温度在很大程度上影响硬盘的寿命。针对数据中心能耗较大的问题，常用的节能策略是尽可能将负载集中到少数的硬盘，让更多的硬盘处于休眠状态，减少高速旋转引起的能耗。但是这种策略会导致高负载硬盘始终处于高温工作状态，引起硬盘故障率急剧上升。因此，研究温度属性对存储设备可靠性的影响具有一定的研究意义。 针对硬盘的可靠性问题，提出一种通过对存储设备温度进行控制来解决存储设备可靠性问题的方法。详细分析了对存储系统温度进行控制所需要实现的具体功能和技术要点。针对存储系统局部过热的问题，引入温度预测和反馈机制，有效提高存储设备的可靠性。同时还使用了阿列纽斯模型来计算当前预测温度下硬盘故障率加速因子和预测后采取关盘措施产生的故障率加速因子，根据这两个故障率加速因子的大小来确定关闭硬盘策略，从而达到温度限制和节能的效果。对于系统中的写操作，使用写离线处理技术，先将写操作写入缓存，再在合适的时候将写操作更新到所有的硬盘，从而有效的减少了写请求对于温度控制策略和节能措施的影响。 测试结果显示，温度控制系统可以较好的控制存储设备的温度，减少硬盘启停频率，有效降低存储设备的故障率，保护存储设备。
[Abstract]:According to the data center server damage, it is found that the temperature affects the hard disk life to a great extent. Aiming at the problem of large energy consumption in data center, the commonly used energy saving strategy is to concentrate the load on a few hard disks as much as possible, so that more hard disks can be kept dormant, and the energy consumption caused by high speed rotation can be reduced. However, this strategy will cause the high-load hard disk to work at high temperature all the time, causing the hard disk failure rate to rise sharply. Therefore, it is significant to study the effect of temperature attribute on storage device reliability. Aiming at the reliability problem of hard disk, this paper presents a method to solve the reliability problem of storage device by controlling the temperature of storage device. Detailed analysis of the storage system temperature control needs to achieve the specific functions and technical points. Aiming at the problem of local overheating of storage system, temperature prediction and feedback mechanism are introduced to effectively improve the reliability of storage devices. At the same time, the Alenius model is used to calculate the acceleration factor of hard disk failure rate at the current predicted temperature and the failure rate acceleration factor caused by the measures taken to close the disk after the prediction. According to the magnitude of these two failure rate acceleration factors, the hard disk closure strategy is determined. Thus, the effect of temperature limitation and energy saving is achieved. For the write operation in the system, the write off line processing technique is used, the write operation is written to the cache first, and then the write operation is updated to all hard disks when appropriate. Therefore, the effect of write request on temperature control strategy and energy saving measures is reduced effectively. The test results show that the temperature control system can better control the temperature of storage devices, reduce the frequency of hard disk startup and stop, effectively reduce the failure rate of storage devices, and protect storage devices.
【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】：TP333.35
本文编号：2151205
[Abstract]:According to the data center server damage, it is found that the temperature affects the hard disk life to a great extent. Aiming at the problem of large energy consumption in data center, the commonly used energy saving strategy is to concentrate the load on a few hard disks as much as possible, so that more hard disks can be kept dormant, and the energy consumption caused by high speed rotation can be reduced. However, this strategy will cause the high-load hard disk to work at high temperature all the time, causing the hard disk failure rate to rise sharply. Therefore, it is significant to study the effect of temperature attribute on storage device reliability. Aiming at the reliability problem of hard disk, this paper presents a method to solve the reliability problem of storage device by controlling the temperature of storage device. Detailed analysis of the storage system temperature control needs to achieve the specific functions and technical points. Aiming at the problem of local overheating of storage system, temperature prediction and feedback mechanism are introduced to effectively improve the reliability of storage devices. At the same time, the Alenius model is used to calculate the acceleration factor of hard disk failure rate at the current predicted temperature and the failure rate acceleration factor caused by the measures taken to close the disk after the prediction. According to the magnitude of these two failure rate acceleration factors, the hard disk closure strategy is determined. Thus, the effect of temperature limitation and energy saving is achieved. For the write operation in the system, the write off line processing technique is used, the write operation is written to the cache first, and then the write operation is updated to all hard disks when appropriate. Therefore, the effect of write request on temperature control strategy and energy saving measures is reduced effectively. The test results show that the temperature control system can better control the temperature of storage devices, reduce the frequency of hard disk startup and stop, effectively reduce the failure rate of storage devices, and protect storage devices.
【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】：TP333.35
【参考文献】
相关期刊论文 前9条
1 郭御风,李琼,刘光明,刘衡竹;虚拟存储技术研究[J];计算机应用研究;2004年02期
2 张金梅;基于模糊推理的智能控制系统的现状和展望[J];科技情报开发与经济;2004年01期
3 王介生,王金城,王伟;基于粒子群算法的PID控制器参数自整定[J];控制与决策;2005年01期
4 王楠;蒋金虎;;存储虚拟化技术研究与比较[J];洛阳师范学院学报;2007年02期
5 雷建龙;;基于单片机的模糊控制器的设计[J];微计算机信息;2006年17期
6 孙海侠;计算机硬盘常见故障及保护措施[J];现代情报;2004年05期
7 吴松,金海;存储虚拟化研究[J];小型微型计算机系统;2003年04期
8 雷志锋;杨少华;黄云;;高功率半导体激光器的可靠性与寿命评价[J];应用光学;2008年01期
9 周红燕;;阿列纽斯方程与电感镇流器的耐久性测试[J];照明工程学报;2006年01期
，本文编号：2151205
本文链接：https://www.wllwen.com/kejilunwen/jisuanjikexuelunwen/2151205.html
