
大数据流式计算环境下的内存节能策略.doc
1页大数据流式计算环境下的内存节能策略蒲勇霖 ;于炯 ;鲁亮 ;廖彬 ;王跃飞【期刊名称】《小型微型计算机系统》 【年(卷),期】2017(038)009【摘要】 针对流式计算系统计算和存储能耗过高的问题 ,改变流式计算中内存的存储状态,提出了一种内存节能策略(ESFM).首先,明确内存中不同状态之间的同 步关系与转换条件 ;其次,通过降低内存活动状态的电压 ,减少处于内存活动状态 的物理节点数量 ,将服务器中的部分内存从活动状态同步成休眠状态 .最后,将计 算后的数据存储于休眠状态的内存空间 .实验证明和理论分析 ,在 24 台普通 PC 机构成的流式计算系统中 ,实施内存节能策略的系统比原系统有效节能约 25.5%.此外,内存节能策略下的性能与能耗的比值为 0.0766tuple/s • J,而原系统性能与能耗的比值为0.0792tuple/s • J.由内可节能策略能够在不影响系统性能的前提下 ,有效降低能耗 .%As the problem of high energy consumption of computing and storage in stream computing system, the Energyefficient Strategy for Memory (ESFM) to change the storage status of computer RAM in stream computing were put forward. First,the synchrony relations and transition condition in the different state of RAMwere definite. Second,by reducing the voltage of memory active state and decreasing the number of physical nodes in the active state of the memory,the part of the server memory from the active state to the hibernation state were changed. Finally,we stored the calculated data in a memory space which is in dormant state. Experimental analysis and。
