I went through two telephonic conversations. In the second call, I was rejected due to negative feedback from the interviewer. However, my answers were correct according to core Hadoop principles.

I work with the Hortonworks distribution, so my answers were based on my experience with Hadoop and core Hadoop.

Microsoft uses the Cloudera distribution, and I believe this is why the interviewer asked platform-specific questions rather than technology-based ones. For example.

How to split a table while using Sqoop?

My answer was --split-by. However, the interviewer was expecting answers like split-by-hash and split-by-mod, which are completely Cloudera platform-dependent.

I had informed the interviewer that I work with the Hortonworks distribution, but they did not consider that.

Again, the interviewer asked me about Parquet, which is a file system purely promoted by Cloudera.

Ecosystems of Hadoop

MapReduce flow

What is Avro and the difference between Avro and text file format?

What are the methods available to split a table while Sqooping?

What is Parquet?