Types of big data
That's actually a question I think more people should ask; and you'd probably get a large variety of answers.
Honestly, speaking, big data covers a plethora of technologies and techniques that are related to handling large amounts of data. So, it's not any specific thing.
People who are "learning big data" generally mean that they are learning the Hadoop ecosystem.
In my mind, having the distributed file system is usually the really important point (though some big data technologies manage a distributed storage system themselves too). Map-reduce is one way of transforming and utilizing data in that file system.
Other ways have been developed over the years which improve on that. For example, Spark is extremely popular for transforming 'big' data in batch or streaming contexts now. Apache Flink is also good in that area.
Tools like Apache Drill can let you query big data in your distributed file system as if it were a SQL database.
Apache HBase (which MapR-DB implements the interface to along with tons of upgraded features) is a distributed data store that works on top of the distributed file system as well. It mimics a large, single, database table and has special storage semantics that work well with the file system. It is great for random reads/writes of big data.
Then there are a plethora of 'big data' databases that may work on or off the distributed file system (e.g. Apache Phoenix can run on top of HBase where as Influx DB and things like that are big data but separate from a hadoop cluster).
Also, streaming technologies often fall under the big data label - e.g. Apache Kafka is a distributed log store that acts as a high volume replacement for traditional messaging technologies in a lot of ways (e.g. it does a lot of what Tibco does or ActiveMQ, but faster and differently).
Anyway; that just scratches the surface. But hopefully that helps you get the idea .
Retrieving data ...