The relation among logical related data is very special with comparing to other style of relation. It can be described by data structure.
In internet and big data time, the data structure of a data instance, the value, is different and variant and specified. Every data structure of data instance is unique, there is no general predefined data type which describes the structure of data instance any more. This makes data user rethink the nature of data structure deeply. Then relational data model and relation database system provide the precisest granularity, aggregate-ignorant, to describe the all details of data structure. This pro or con would make the database designer crazy and the designing complex and more easy proven error. The NoSQL database system have provided more data model with levels of granularity, such as no type limitation in cell, to describe the structure of data.
How does the different data structure make the different data type? The nature of data type is about not only the relationship among data elements, such as the order, but also the relationship between the data elements and the unity of object type. Abstract data structure is a way to express different relationship among data, different structure means different relation expressing. However, the data structure is just static, determining the properties or nature of data, not the objective dynamic relation or operations which can change the state of data.
The type of data elements(node) and the order or space location of composing elements(hierarchy) make a different data structure or type. The information of order and type of elements is significant.
The information of data stored in a database includes two independent aspects, the location information and the type information. Location information should not maintain the type information. But the relation data model in relational database intertwines the location information and the type information together, that means the column in the table of database has two kind information of data, the location and the type. This is the trade-off of relational da