02 Graph Databases

Motivations

The table based structure of relational databases makes it hard to represent relationships between rows in the same table,and moreover whenever someone needs to find a relationship between records of different tables the db has to perform a JOIN operation, which is usually very expensive.

So for the use cases in which relationships are the most important feature of our data (e.g. social network friendships) it would be best to go with a technology who can implement relationships in a native and efficient way,and that’s where graph dbs come in.

Definition

"Database that uses graph structures with nodes, edges and properties to store data"

Graphs provides index-free adjacency: every node is a pointer to its adjacent element. Edges hold most of the important information.

In these kind of storage systems data are represented as entities connected by information rich relations,just like in a real graph.

Query model

The typical query based approach for managing data in relational databases isn’t a good fit for graph dbs,in these cases it’s preferred a different approach called Graph Matching, or more in general Pattern Matching. In these technique the user specifies a particular shape or structure he wants to find in the graph, and the system searches for and returns all the subgraphs that match that request.