Unifying Big Data Batch and Real-Time Streaming with Apache Flink

Note: Please arrive as early as 5pm for pizza, beverages, and networking. The presentation will begin at 6.

Apache Flink continues on the ideas from Hadoop but adds optimization and transformation mechanisms from distributed databases and parallel collections. Flink runs on top of HDFS and YARN, but the execution is optimized, in a similar way as the relational databases optimize SQL. The execution model is based on a memory management scheme that favors in-memory processing, but then gracefully degrades to disk when necessary. The same engine supports both batch and true streaming. Flink has a very elegant Scala based API, which highly resembles using Scala collection libraries.

Join us in exploring the Apache Flink’s state of the art data processing capabilities, end user benefits and new Big Data algorithms.

Outline:

• Flink Architecture
• Using Flink
• Flink Execution and Internals
• Batch and Stream Processing
• Hadoop Compatibility

Dr. Vladimir Bacvanski

Dr. Vladimir Bacvanski has over two decades of engineering experience with mission critical and distributed enterprise systems and data technologies. Vladimir has helped a number of companies including US Treasury, Federal Reserve Bank, US Navy, IBM, Dell, Hewlett Packard, JP Morgan Chase, General Electric, BAE Systems, AMD, and others to select, transition to, and apply new software and data technologies. Vladimir is published worldwide and is a keynote speaker, session chair, and workshop organizer at leading industry events. As a founder of SciSpike (http://www.scispike.com/), Vladimir is focusing on Big Data technologies and highly scalable reactive software architectures with node.js and Scala. Vladimir is the author of the new O'Reilly course on Big Data and NoSQL (http://shop.oreilly.com/product/0636920040804.do).