The first alpha of Diaspora, released a while back, was using MongoDB and that brought both project some press.
Social data is relational. […] amazingly enough, document databases aren’t great at storing relational data
That’s a very interesting remark.
Leaving aside for a second the size of social data which some would say poses questions about the cost effectiveness of using relational databases, I think this is just another example of confusing relationships for relational model.
Social data is fundamentally unstructured and that makes it a good fit for schemaless storage. Even those that are using RDBMSes for storing social data are bending them to fit unstructured inside a schema constraint storage.
On the other hand, social data keeps its value if existing relationships are preserved.
So, is there a solution that would allow one to store unstructured data showing strong relationships? And I’d say the answer is obviously graph databases.
Update: Sarah Mei mentioned on Twitter that she actually meant relational, but didn’t provide any other details.
Original title and link: Diaspora Drops MongoDB for MySQL or Social Data and Graph Databases (NoSQL databases © myNoSQL)