The Fifth Elephant 2015

A conference on data, machine learning, and distributed and parallel computing

Deprecating MapReduce Patterns with Apache Spark

Submitted by Rahul Kavale (@rahulkavale) on Thursday, 7 May 2015

Section: Full Talk Technical level: Intermediate


Live coding demostration to show how Apache Spark can solve non trivial problems, and hence deprecating some of the established patterns of MapReduce, with consice code, giving us significant performance gain, and developer friendly programming model also keeping other sweetness of MapReduce wolrd intact.


Hadoop MapReduce programming model has evolved into having its own patterns for solving problems. This talk is trying to eloborate how those patterns can be replaced with simpler equivalent solutions with use of Apache Spark, and of course with keeping other sweetness of existing MapReduce world with advantage of performance boost. I will be using some some non trivial examples for the this with live coding for the same.

Speaker bio

Rahul is an Application developer with Thoughtworks. Rahul has worked with technologies ranging from Scala, Java, Ruby. He has experience in builiding web applications to solving big data problems. Rahul has special interest in Scala, Apache Spark. He loves functional programming.



  • Shashi Gowda (@g0wda) 4 years ago

    The content of your other proposal seems redundant. Why not combine these two talks?

    • Rahul Kavale (@rahulkavale) Proposer 4 years ago

      Hi Shashi, this talk is a live coding demo, in which I will try to elaborate on how to implement typical problems with Spark, for which we have some patterns in MapReduce world. This talk aims to focus Spark via live coding examples and not go in depth for any underlying details about its execution and internals while the other one is reverse of this. Hence this one is different from the another one I have submitted.

Login with Twitter or Google to leave a comment