There is option to use Sesame to be used both as querying the RDF data using SPARQL and also the Native database storage, but there is also option to use the Sesame only as the querying language and use the MongoDB/MySQL or some other database storage.
Which one should be in which scenario ? What are benefits and disadvantage of using Sesame as a Native storage ?
First of all: there is no option to use MongoDB in combination with Sesame (at least none that I know of). There is a legacy feature to use Sesame with MySQL or PostgreSQL, but this has been deprecated since Sesame 2.7 and is not a recommended setup in any scenario.
Sesame is primarily an API and RDF toolkit for which other parties provide database solutions (both commerciel and open souce), though it comes with two bundled database implemetations, the Memory Store and the Native Store. The memory store does what it says on the tin: it's fast but keeps everything in main memory.
The native store is a proper persistent database, and is designed to perform well for small to medium-sized data sets, order of 100-200 million triples. Beyond that, you need to look into other Sesame API-compatible solutions, such as OWLIM, BigData, CumulusRDF, or Stardog (to name a few).