Why Java Code in Hadoop uses own Data Types instead of basic Data types

Question

I have just started writing MapReduce codes in Hadoop and I’m not sure why Hadoop uses it’s own data types like BooleanWritable, ByteWritable, IntWritable, LongWritable instead of using basic Java data types. Can anyone help me in understanding this?

Answer 1

Hadoop provides us Writable interface based data types for serialization & de-serialization in MapReduce computation. Using Writables like BooleanWritable, ByteWritable, IntWritable, LongWritable provides efficiency in computation.

They also give us compact representation as they do not store the serialized data format. And it is already known how to de-serialize the data. This is the reason why, we are using Writable interface instead of Java serialization.