|
|
【答案】应助回帖
 商家已经主动声明此回帖可能含有宣传内容
★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ...
yuxintian: 金币+90, 翻译EPI+1, ★★★★★最佳答案, 90 2015-10-16 18:44:28
MapReduce计算框架 的 核心思想 是把 对大规模数据集 的操作,分发到 一个 主节点
管理下 的各分节点 共同完成, 然后 通过 整合 各个分节点 产生的 中间结果，得到
最终结果。图1 显示了 MapReduce计算框架 的 计算流程。可以看出 整个处理过程 被
MapReduce计算框架 高度地 抽象为 两个函数：Map和Reduce。Map函数 主要负责 将任
务 分解成为 多个 并行执行的 任务，Reduce函数 主要负责将 这些 并行执行的任务
所产生的 结果 汇总处理。至于 在并行编程中 会遇到的 其他 种种复杂问题、如分布
式存储、工作调度、负载均衡、网络通信等 都由 MapReduce计算框架 负责处理，使编
写 和 运行 用于 处理海量数据 的 应用程序 更加容易。
The core thought of MapReduce computing framework is to distribute the large-scale data processing operation into various partial node managed under a main node, then through integrating the intermediate results resulted from each partial node, a final result can be obtained. Fig.1 shows the computing flow of MapReduce computing framework, from which tt can be seen that the whole process is abstracted by MapReduce computing framework into two functions:Map and Reduce. The Map function is responsible for decomposing the task into mult-tasks in parallel, while the Reduce function is mainly used to collect and process the paralleled tasks. As for various complex problem which may be encountered in multiple programming such as distributed storage, job scheduling, load balance, and network communication, MapReduce computing framework will solely take care of them, so as to make it easier to write and operate application programs that are used for massive data processing. |
|
回复此楼