Basically, it involves taking massive amounts of data and using computer algorithms to find patterns and trends. So at the end of the day machine learning is nothing without data. The most preferable Storage system to be used with machine learning is Hadoop HDFS. Hadoop is an open-source tool for storing and data processing in a distributed environment. It is designed to scale up from single servers to thousands of machines, each offering local computation and large storage.

But the question arises that when there is no need to store millions of records then why shouldn’t we use NOSQL. Like Hadoop, NoSQL is developed for distributed and parallel computing. The main difference is Hadoop is not a database system but is a software ecosystem that allows for massively parallel computing. But, NoSQL is a database designed especially for unstructured data and is very efficient and flexible by nature. So at the end of the day the question arises especially for young startups, and MSME Sector why we need to pay (both in terms of money, and knowledge) for distributed infrastructure such as Hadoop when we can go for NoSQL that can be setup on local machines.