In RDBMS, goals of Normalization:
Free the database of modification anomalies
Minimize re-design when extending
Avoid bias toward any particular access pattern
First step involves avoiding redundancy etc...
As data is normalised and divided in multiple tables, join operations are required to satisfy queries placed by application. join operation is costly.
With relational model, there is an issue of object mismatch i.e., gap between real world(application access) and relational world.
On the contrast, Document databases(like MongoDB) do not have object mismatch issue.
The single most important factor in designing the database schema with document database, is about, matching the data access patterns of your application. So, database design is pretty easy.
They also support transactions, now
De-normalisation issue still gets carried in Document database, unless you link data among collections and map/reduce on application access
Why document oriented database is yet to replace relational model database?