是一个文件系统，用于存储文件，通过统一的命名空间目录树来定位文件； 其次，它是分布式的，由很多服务器联合起来实现其功能，集群中的服务器有各自的角色。

HDFS是一个主/从（Mater/Slave）体系结构，由三部分组成： NameNode 和 DataNode 以及 SecondaryNamenode： NameNode 负责管理整个文件系统的元数据，以及每一个路径（文件）所对应的数据块信息。 DataNode 负责管理用户的文件数据块，每一个数据块都可以在多个 DataNode 上存储多个副本，默认为3个。 Secondary NameNode 用来监控 HDFS 状态的辅助后台程序，每隔一段时间获取 HDFS 元数据的快照。最主要作用是辅助 NameNode 管理元数据信息。

研究者July在计算机和机器学习领域的水平怎么样？

如何看待阿里9000万欧元收购Apache Flink母公司Data Artisans？

Apache Flink和Apache Spark发展前景分别怎样？

大数据开发像Hadoop、Spark需要很强的数学和算法功底吗？

如何看待R大加入大数据公司Databricks？

Spark 千万级用户相似度计算？

如何进入大数据领域，学习路线是什么？

大数据公司 Splunk 和 Cloudera 的核心竞争力在哪里？

从百万级别数据的分析角度，Mysql，Mongodb，Hbase如何选择？

大数据分析/挖掘尽管重要，产品的信息化也十分重要，企业需要一个从大数据到大产品的进程。

2 HOMER     可再生能源互补发电优化建模 ( Hybrid Optimization Model for Electric Renewable，HOMER) 是由 NREL 资助开发的可再生能源混合发电经济-技术-环境优化分析计算模型，主要针对小功率可再生能源发电系统结合常规能源发电系统形成的混合发电系统进行优化。     HOMER 的优点在于其灵活的系统建模能力，能够对多种可再生能源、发电技术进行建模仿真，储能模型考虑了飞轮、蓄电池、液流电池以及氢储能。HOMER能够对并网型和独立型微电网系统进行建模仿真，支持基于全年 8 760 h 能量平衡仿真的系统容量优化以及参数灵敏度分析。其应用范围广泛，适用于不同规模的系统，目前已在城市、海岛、村庄、社区、住宅等规模下的可再生能源规划及电网优化设计中得到应用。此外，HOMER 还能提供不同系统配置下详细的经济分析结果，但不足是作为能源规划分析软件，没有对网络进行建模。

We’re in the middle of a data revolution. Business processes everywhere are becoming digitized. Firms like Walmart and Target know exactly what you search for and what you end up buying. Other firms like ADP, Mastercard and FedEx are intimately involved in payrolls, transactions, delivery and every other stage of the commercial pipeline. And every single action that these firms take is recorded and stored for analysis. Human interactions are also becoming digitized. Social networks, instant messaging and web search paint a dynamic, real-time picture of what people are interested in and who they’re talking to. Again, every single action is recorded and stored for posterity. Smartphones are ubiquitous. This means an accurate location sensor, audio recorder, still/video camera, radio transponder and internet connection in every pocket. Almost no part of the world is outside the limits of cellular coverage. Cars and trucks now have embedded sensors, tracking position, velocity, traffic and much more. Satellites and GPS have gone from the preserve of the few (military) to the plaything of the many; imagery and position data are today a public good. As a result of these technological innovations, we are swimming in a sea of data. Yet this data would be meaningless if it weren’t for another, parallel advancement in the area of computation. Thanks to the relentless progress of Moore’s Law, we have the bandwidth to capture all this data, the memory to store it and the cycles to analyze it and extract commercial value from it. This capacity has transformed industries everywhere.

And companies will evolve to make use of all this data. Today, for every Amazon or Target that runs a rigorous program of data collection, analysis and action, there are dozens of companies that still operate by old-school rules. As time goes by, these firms will either adapt or become extinct. Either way, the future belongs to businesses that embrace the data revolution. This means that the amount of business data available to analysts will only continue to grow.