Chapter 7. Streaming and Real-time Analysis – Storm and Spark

As we have already discussed about Hadoop being a Batch processing system and some data source types that varies in their velocity or rate, volume of data. Many system especially machines generates a lot of data consistently, they need to process such high volume data to maintain quality and avoid heavy loss and thus the need for Stream processing has emerged. To design systems that are built as Lambda implementation, which are Batch as well as Stream processing systems, We should have combination of different environment that can integrate with each other to process the data and quite obviously which increases the complexity of designing the system. Streaming data is complex to store, analyze, process, and maintain. Prior to version 2.x, Hadoop was only a Batch processing system, and after the emergence of YARN and other frameworks and the integration of those frameworks with YARN, Hadoop can be designed for...