How to migrate data from MongoDB to SQL-Server?

Of course it's possible, but you will need to find a way to force the flexibility of a document db like MongoDB into a RDBMS like SQL Server.

It means that you need to define how you want to handle missing fields (will it be a NULL in the db column? or a default value?) and other things that usually don't fit well in a relational database.

Said do, you can use an ETL tool able to connect to both databases, SSIS can be an example if you want to stay in the MicroSoft world (you can check this Importing MongoDB Data Using SSIS 2012 to have an idea) or you can go for an open source tool like Talend Big Data Integration which has a connector to MongoDB (and of course to SQL Server).

There is a tool provided by MongoDB called mongoexport and it's capable of exporting csv files. These csv files can be easily imported into MySQL. Good luck!

There is no way to directly move data from MongoDB to SQL Server. Because MongoDB data is non-relational, any such movement must involve defining a target relational data model in SQL Server, and then developing a transformation that can take the data in MongoDB and transform it into the target data model.

Most ETL tools such as Kettle or Talend can help you with this process, or if you're a glutton for punishment, you can just write gobs of code.

Keep in mind that if you need this transformation process to be online, or applied more than once, you may need to tweak it for any small changes in the structure or types of the data stored in MongoDB. As an example, if a developer adds a new field to a document inside a collection, your ETL process will need rethinking (possibly new data model, new transformation process, etc.).

If you are not sold on SQL Server, I'd suggest you consider Postgres, because there is a widely-used open source tool called MoSQL that has been developed expressly for the purpose of syncing a Postgres database with a MongoDB database. It's primarily used for reporting purposes (getting data out of MongoDB and into an RDBMS so one can layer analytical or reporting tools on top).

MoSQL enjoys wide adoption and is well supported, and for badly tortured data, you always have the option of using the Postgres JSON data type, which is not supported by any analytics or reporting tools, but at least allows you to directly query the data in Postgres. Also, and now my own personal bias is showing through, Postgres is 100% open source, while SQL Server is 100% closed source. :-)

Finally, if you are only extracting the data from MongoDB to make analytics or reporting easier, you should consider SlamData, an open source project I started last year that makes it possible to execute ANSI SQL on MongoDB, using 100% in-database execution (it's basically a SQL-to-MongoDB API compiler). Most people using the project seem to be using it for analytics or reporting use cases. The advantage is that it works with the data as it is, so you don't have to perform ETL, and of course it's always up to date because it runs directly on MongoDB. A disadvantage is that no one has yet built an ODBC / JDBC driver for it, so you can't directly connect BI tools to SlamData.

Good luck!