Is data sent to some other location to be processed in pig

Question

We load the data from HDFS, process the data in Pig and store the result back to HDFS. The basic principle of Hadoop is - bringing processing to data rather than taking data to process. So is this high-level implementation is just to make us understand - not the actual way how it works?

Answer 1

With pig, we are not taking data anywhere. The data that we have is in hdfs, the processing that is taking place is happening across the hadoop cluster itself with the help of map reduce. Pig is not a processing place. It is just doing the work of semantics that is the pig compiler's duty is to transform a pig latin program into a series of Java Map Reduce task. Pig helps by just allowing the user to focus on semantics.