发明名称 Method and apparatus for storing data
摘要 Embodiments of the present invention provide a method and an apparatus for storing data, which relate to the field of data processing. In the present invention, a current device is divided into different load modes in the process of service processing, and manners of storing various data in a Cache are dynamically adjusted, so that nodes with different characteristics in the current device may control operations on the Cache, thus achieving lower power consumption and optimum performance of a large-capacity system under a heavy load.
申请公布号 US9208085(B2) 申请公布日期 2015.12.08
申请号 US201213730705 申请日期 2012.12.28
申请人 Huawei Technologies Co., Ltd. 发明人 Cheng Chuanning
分类号 G06F12/08 主分类号 G06F12/08
代理机构 代理人
主权项 1. A method for storing data, the method comprising: judging a load mode of a current device; when the load mode is a first load mode, storing, in a Cache, all data being processed by the current device; when the load mode is a second load mode, storing, in the Cache, data being processed by the current device and having a use frequency exceeding a preset frequency, and storing, in a lower level memory of the Cache, data having a use frequency not exceeding the preset frequency; and wherein judging the load mode of the current device comprises: obtaining ingress node actual bandwidth traffic R, the number of tasks in a task queue, a processing delay T, a Cache hit rate and the number of times of receiving a back-pressure signal or a back-pressure message in the current device through monitoring, andjudging the load mode of the current device according to the ingress node actual bandwidth traffic R, the number of tasks in the task queue, the processing delay T, the Cache hit rate and the number of times of receiving the back-pressure signal or the back-pressure message in the current device, wherein obtaining the ingress node actual bandwidth traffic R, the number of the tasks in the task queue, the processing delay T, the Cache hit rate and the number of times of receiving the back-pressure signal or the back-pressure message in the current device through monitoring comprises: setting up a first monitoring point at an ingress node of the current device, and monitoring the first monitoring point in a preset first time period to obtain the ingress node actual bandwidth traffic R;setting up a second monitoring point at the Cache of the current device, and monitoring the second monitoring point in the preset first time period to obtain the Cache hit rate;setting up a third monitoring point at a processing node of the current device, and monitoring the third monitoring point in the preset first time period to obtain the number of tasks to be processed in the processing node;setting up a fourth monitoring point at an egress node of the current device, and monitoring the fourth monitoring point to obtain the number of tasks to be processed at the egress node;counting the number of tasks in the task queue in the preset first time period, wherein the number of the tasks in the task queue is a sum of the number of the tasks to be processed in the processing node and the number of the tasks to be processed at the egress node; andobtaining, through calculation, the processing delay T for processing data by the current device, wherein the current device at least comprises the ingress node, the egress node and the processing node configured to perform service processing on the data.
地址 Shenzhen CN