RDBMS were never relational!
As per my knowledge, Oracle and a few database providers ruled the world for more than three decades as a relational database. The ground requirement was not to store/retrieve the data but to build information in a semantic way which can be useful to create knowledge for business and other useful domains. The shocking ground reality of all these databases is, Not even a single database is relational. So there was no question of build the knowledge for semantic analysis. These databases are capable to store all the data in a very well manner but not the relationship in an intuitive manner that can be mapped as is with any of the real-world domains.
The user was diverted and was forced to use normalization to create primary, foreign key, and map that to create a so-called hardcoded data relationship. Obviously, it would create so many constraints and limitations in terms of expected information, response time, knowledge insight to know what we know and what we don’t know. Yes, we were privileged by using some magic keywords like join, rollup, the cube to retrieve so-called relational information. Apart from a normal(single hop to a certain hop) query, behind these magical words, compute resources always be on fire. If we think to do any operation involve 3–4 or more hop then don’t even think about response time as for such kind of use case involve recursive query with intense resources.
So to say, I am not challenging any databases neither my intention is to start a technology war. My whole intention is just to highlight that what is the meaning of relational data and what we used as relational databases. As the world is semantic and everything is connected with every other thing having a relationship. So the world is overloaded with data but still, everyone is struggling to generate information from data, knowledge from information, and smart knowledge from knowledge.
The system which is capable to store real-world relationship as is at the physical layer can help to build the smart knowledge. so far we see only graph databases having these capabilities. They store a complex network of data from any domain however their storage mechanism is not complicated, neither it depends on the size of the dataset and type of network. There are multiple types of graph models. These are Directed-edge labeled aka RDF, Graph data sets, and property graph.
We start designing any graph model by defining nodes and relationships. To name these nodes and define properties there are W3C published standards known as OWL. Using this we create nodes and define the relationship. For CRUD operation, many graph language has been proposed by the relevant tools ..few of them are Sparql, Cypher, aspen. The scope is much wider and a lot of research is happening in this area. With Google, Facebook, and ..(a big list) started using the graph datasets to represent data in a semantic manner without any known limitation, leveraging the capability of graph database will change the way we see data, we use the data for almost any and every use-case, wherever applicable.
