Originally posted on AnalyticBridge
When most technical professionals think of Big Data analytics today, they think of Hadoop. But there are many cutting-edge applications that Hadoop isn't well suited for, especially real-time analytics and contexts requiring the use of iterative machine learning algorithms. Fortunately, several powerful new technologies have been developed specifically for use cases such as these. Big Data Analytics Beyond Hadoop is the first guide specifically designed to introduce these technologies and demonstrate their use in detail. An indispensable resource for data scientists and others who must scale traditional analytics tools and applications to Big Data, it illuminates these new alternatives at every level, from architecture all the way down to code. Dr. Vijay Srinivas Agneeswaran shows how to evaluate and choose the right tools, and then reengineer your solutions and products to work far more effectively in Big Data environments. Agneeswaran explains the Berkeley Data Analysis Stack (BDAS) in detail, including its motivation, design, architecture, Mesos cluster management, and the analysis of both performance and accuracy. He presents realistic use cases and up-to-date example code for:
- Spark, the next generation in-memory computing technology from UC Berkeley
- Storm, the parallel real-time Big Data analytics technology from Twitter
- GraphLab, the next-generation graph processing paradigm from CMU and the University of Washington (with comparisons to alternatives such as Pregel and Piccolo)
Agneeswaran offers architectural and design guidance and code sketches for scaling machine learning algorithms to Big Data, and then realizing them in real-time. He concludes by previewing emerging trends, including real-time video analytics, SDNs, and even Big Data governance, security, and privacy issues. To position you for tomorrow's advances, he identifies intriguing startups and new research possibilities, including BDAS extensions and cutting-edge model-driven analytics.