There is absolutely a clear advantage when it comes to prototyping and rapid development. Specific example: Ever built a crawler that harvested hundreds of gigabytes of raw data and then realized you needed to make a schema change later? I don't want to take the time to think about every single use case and every single column I'll need and their datatypes. I just want to move on so that I can start doing things with the data.
I still think of data as rows in MongoDB, but the lack of a fixed schema makes life seriously easy. That means my schema is defined in the application (and only the application) and is subject to versioning. Want an extra column? Just add it to your application and you're done. No need to promote a slave and cycle through while waiting hours upon hours for each table to build.
I still think of data as rows in MongoDB, but the lack of a fixed schema makes life seriously easy. That means my schema is defined in the application (and only the application) and is subject to versioning. Want an extra column? Just add it to your application and you're done. No need to promote a slave and cycle through while waiting hours upon hours for each table to build.